feat: increase main layout max-width to enhance content display

Co-authored-by: Copilot <copilot@github.com>

.gitignore

@@ -45,4 +45,8 @@ Thumbs.db
 
 # instructions
 .github/instructions/
+
+# Logs and generated data
+logs/
+data/
 backend/data/

README.md

@@ -1,7 +1,16 @@
-# AI
+# All You Can GET AI
 
 A multi-modal AI web application. Users can choose between different AI models for text generation, text-to-image, text-to-video, and image-to-video generation, powered by [openrouter.ai](https://openrouter.ai).
 
+Key features:
+
+- Multi-modal AI generation (text, images, videos)
+- User authentication and role-based access control
+- Admin dashboard for managing users, models, and video jobs
+- Gallery for viewing generated images and videos
+- Chat interface with message history
+- Image upload and preview functionality
+
 ## Components
 
 | Component | Technology        | Description                                              |
@@ -31,49 +40,98 @@ python -m venv .venv
 # Linux/macOS
 source .venv/bin/activate
 
-# Install dependencies
+# Install core dependencies
 pip install -r requirements.txt
 
-# Copy and fill in environment variables
+# Install development dependencies
+pip install -r backend/requirements-dev.txt
+pip install -r frontend/requirements-dev.txt
+
+# Copy environment variables file
 cp .env.example .env
+
+# Edit .env file and add your OpenRouter API key and configure other settings
+nano .env
 ```
 
-### Running the backend
+### Running the application locally
+
+#### Backend (FastAPI + Uvicorn)
 
 ```bash
 cd backend
 uvicorn app.main:app --reload --port 12015
 ```
 
-### Running the frontend
+#### Frontend (Flask)
 
 ```bash
 cd frontend
-flask --app app.main run --port 12016
+flask --app app.main run --port 12016 --debug
 ```
 
 ### Running tests
 
 ```bash
+# Run all tests
 pytest
+
+# Run backend tests only
+pytest backend/tests/
+
+# Run frontend tests only
+pytest frontend/tests/
 ```
 
+### Available Environment Variables
+
+| Variable             | Description                 | Default             |
+| -------------------- | --------------------------- | ------------------- |
+| `OPENROUTER_API_KEY` | Your OpenRouter API key     | _Required_          |
+| `ADMIN_EMAIL`        | Default admin user email    | `ai@allucanget.biz` |
+| `ADMIN_PASSWORD`     | Default admin user password | `admin123`          |
+| `DATABASE_URL`       | DuckDB database path        | `../data/app.db`    |
+
+## Default admin user
+
+On first startup a default admin account is created:
+
+| Field    | Value               |
+| -------- | ------------------- |
+| Email    | `ai@allucanget.biz` |
+| Password | `admin123`          |
+| Role     | `admin`             |
+
+Override via environment variables `ADMIN_EMAIL` and `ADMIN_PASSWORD` before first run.
+
+## Deployment
+
+Deployed on [Coolify](https://coolify.io) using Nixpacks. See [docs/deployment/coolify.md](docs/deployment/coolify.md) for full instructions.
+
 ## Project Structure
 
 ```txt
 backend/                  FastAPI backend
   app/
-    routers/      API route handlers
-    services/     Business logic
-    models/       Pydantic models
-  tests/
+    __init__.py          Package initialization
+    db.py                Database connection and operations
+    dependencies.py      Dependency injection
+    main.py              FastAPI application entrypoint
+    models/              Pydantic and database models
+    routers/             API route handlers (auth, users, admin, generate, gallery)
+    services/            Business logic for AI generation, users, admin, etc.
+  tests/                  Backend test suite
 frontend/                 Flask frontend
   app/
+    __init__.py          Package initialization
+    main.py              Flask application entrypoint
     templates/           Jinja2 HTML templates
     static/              CSS, JS, images
-  tests/
-data/             DuckDB database files (gitignored)
-docs/             Architecture documentation
+  tests/                  Frontend test suite
+data/                     DuckDB database files, uploaded media, and generated content
+logs/                     Application logs
+docs/                     Architecture documentation (arc42 template)
+nginx/                    Nginx configuration for Coolify deployment
 ```
 
 ## Documentation

backend/app/db.py

@@ -64,3 +64,86 @@ def _run_migrations(conn: duckdb.DuckDBPyConnection) -> None:
             revoked     BOOLEAN   DEFAULT false
         )
     """)
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS uploaded_images (
+            id           UUID      DEFAULT uuid() PRIMARY KEY,
+            user_id      UUID      NOT NULL,
+            filename     VARCHAR   NOT NULL,
+            content_type VARCHAR   NOT NULL,
+            file_path    VARCHAR   NOT NULL,
+            size_bytes   BIGINT    NOT NULL,
+            created_at   TIMESTAMP DEFAULT now()
+        )
+    """)
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS models_cache (
+            id             UUID      DEFAULT uuid() PRIMARY KEY,
+            model_id       VARCHAR   NOT NULL UNIQUE,
+            name           VARCHAR   NOT NULL,
+            modality       VARCHAR   NOT NULL,
+            context_length BIGINT,
+            pricing        JSON,
+            fetched_at     TIMESTAMP NOT NULL
+        )
+    """)
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS generated_images (
+            id          UUID      DEFAULT uuid() PRIMARY KEY,
+            user_id     UUID      NOT NULL,
+            model_id    VARCHAR   NOT NULL,
+            prompt      VARCHAR   NOT NULL,
+            image_data  VARCHAR   NOT NULL,
+            created_at  TIMESTAMP DEFAULT now()
+        )
+    """)
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS generated_videos (
+            id          UUID      DEFAULT uuid() PRIMARY KEY,
+            user_id     UUID      NOT NULL,
+            job_id      VARCHAR   NOT NULL,
+            model_id    VARCHAR   NOT NULL,
+            prompt      VARCHAR   NOT NULL,
+            polling_url VARCHAR,
+            status      VARCHAR   NOT NULL DEFAULT 'pending',
+            video_url   VARCHAR,
+            created_at  TIMESTAMP DEFAULT now(),
+            updated_at  TIMESTAMP DEFAULT now()
+        )
+    """)
+    # Migration: add output_modalities column if absent (stores JSON array string)
+    conn.execute("""
+        ALTER TABLE models_cache ADD COLUMN IF NOT EXISTS output_modalities VARCHAR
+    """)
+    # Migration: add video job request params + generation type
+    conn.execute("""
+        ALTER TABLE generated_videos ADD COLUMN IF NOT EXISTS request_params VARCHAR
+    """)
+    conn.execute("""
+        ALTER TABLE generated_videos ADD COLUMN IF NOT EXISTS generation_type VARCHAR DEFAULT 'text_to_video'
+    """)
+    conn.execute("""
+        ALTER TABLE generated_videos ADD COLUMN IF NOT EXISTS error VARCHAR
+    """)
+    _seed_admin(conn)
+
+
+def _seed_admin(conn: duckdb.DuckDBPyConnection) -> None:
+    """Insert the default admin user if it doesn't already exist."""
+    from passlib.context import CryptContext
+    _pwd = CryptContext(schemes=["bcrypt"], deprecated="auto")
+
+    email = os.getenv("ADMIN_EMAIL", "ai@allucanget.biz")
+    password = os.getenv("ADMIN_PASSWORD", "admin123")
+
+    existing = conn.execute(
+        "SELECT id FROM users WHERE email = ?", [email]
+    ).fetchone()
+    if existing is None:
+        password_hash = _pwd.hash(password)
+        conn.execute(
+            """
+            INSERT INTO users (email, password_hash, role)
+            VALUES (?, ?, 'admin')
+            """,
+            [email, password_hash],
+        )

backend/app/main.py

@@ -1,9 +1,13 @@
-from .routers import auth as auth_router
-from .routers import users as users_router
-from .routers import admin as admin_router
-from .routers import ai as ai_router
-from .routers import generate as generate_router
-from .db import close_db, init_db
+from .routers import auth
+from .routers import users
+from .routers import admin
+from .routers import ai
+from .routers import generate
+from .routers import images
+from .routers import models
+from .db import close_db, get_conn, get_write_lock, init_db
+from .services.video_worker import run_worker
+import asyncio
 import os
 from contextlib import asynccontextmanager
 
@@ -17,12 +21,18 @@ load_dotenv()
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     init_db()
+    worker_task = asyncio.create_task(run_worker(get_conn(), get_write_lock()))
     yield
+    worker_task.cancel()
+    try:
+        await worker_task
+    except asyncio.CancelledError:
+        pass
     close_db()
 
 
 app = FastAPI(
-    title="AI Allucanget Biz API",
+    title="All You Can GET AI Biz API",
     description="Multi-modal AI generation API powered by openrouter.ai",
     version="0.1.0",
     lifespan=lifespan,
@@ -36,11 +46,13 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-app.include_router(auth_router.router)
-app.include_router(users_router.router)
-app.include_router(admin_router.router)
-app.include_router(ai_router.router)
-app.include_router(generate_router.router)
+app.include_router(auth.router)
+app.include_router(users.router)
+app.include_router(admin.router)
+app.include_router(ai.router)
+app.include_router(generate.router)
+app.include_router(images.router)
+app.include_router(models.router)
 
 
 @app.get("/health", tags=["health"])

backend/app/models/ai.py

@@ -33,8 +33,9 @@ class ModelInfo(BaseModel):
 
 class TextRequest(BaseModel):
     model: str
-    prompt: str
+    prompt: str = ""
     system_prompt: str | None = None
+    messages: list[ChatMessage] | None = None
     temperature: float = 0.7
     max_tokens: int = 1024
 
@@ -61,6 +62,7 @@ class ImageResult(BaseModel):
     url: str | None = None
     b64_json: str | None = None
     revised_prompt: str | None = None
+    image_id: str | None = None  # UUID of stored row in generated_images
 
 
 class ImageResponse(BaseModel):
@@ -89,7 +91,8 @@ class VideoFromImageRequest(BaseModel):
 
 
 class VideoResponse(BaseModel):
-    id: str
+    id: str  # This is the job_id from the provider
+    db_id: str | None = None  # This is the UUID from our generated_videos table
     model: str
     status: str  # "queued" | "processing" | "completed" | "failed"
     polling_url: str | None = None

backend/app/routers/admin.py

@@ -1,10 +1,13 @@
 """Admin router: operational endpoints for application management."""
-from datetime import datetime, timezone
+from datetime import datetime, timedelta, timezone
+from typing import Any
 
 from fastapi import APIRouter, Depends
 
 from ..db import get_conn, get_write_lock
 from ..dependencies import require_admin
+from ..services import models as models_service
+from ..services.models import mark_timed_out_video_jobs
 
 router = APIRouter(prefix="/admin", tags=["admin"])
 
@@ -13,16 +16,23 @@ router = APIRouter(prefix="/admin", tags=["admin"])
 async def get_stats(_: dict = Depends(require_admin)) -> dict:
     """Return aggregate statistics: user counts and token counts."""
     conn = get_conn()
-    total_users = conn.execute("SELECT COUNT(*) FROM users").fetchone()[0]
-    users_by_role = conn.execute(
-        "SELECT role, COUNT(*) FROM users GROUP BY role ORDER BY role"
-    ).fetchall()
-    total_tokens = conn.execute(
-        "SELECT COUNT(*) FROM refresh_tokens").fetchone()[0]
-    active_tokens = conn.execute(
-        "SELECT COUNT(*) FROM refresh_tokens WHERE revoked = false AND expires_at > ?",
-        [datetime.now(timezone.utc)],
-    ).fetchone()[0]
+    sql_user_count = "SELECT COUNT(*) FROM users"
+    sql_user_counts = "SELECT role, COUNT(*) FROM users GROUP BY role ORDER BY role"
+    sql_token_count = "SELECT COUNT(*) FROM refresh_tokens"
+    sql_tokens_active = "SELECT COUNT(*) FROM refresh_tokens WHERE revoked = false AND expires_at > ?"
+    now = datetime.now(timezone.utc)
+
+    total_users_row = conn.execute(sql_user_count).fetchone()
+    total_users = total_users_row[0] if total_users_row else 0
+
+    users_by_role = conn.execute(sql_user_counts).fetchall()
+
+    total_tokens_row = conn.execute(sql_token_count).fetchone()
+    total_tokens = total_tokens_row[0] if total_tokens_row else 0
+
+    active_tokens_row = conn.execute(sql_tokens_active, [now]).fetchone()
+    active_tokens = active_tokens_row[0] if active_tokens_row else 0
+
     return {
         "users": {
             "total": total_users,
@@ -40,7 +50,8 @@ async def get_stats(_: dict = Depends(require_admin)) -> dict:
 async def db_health(_: dict = Depends(require_admin)) -> dict:
     """Verify DuckDB is reachable."""
     conn = get_conn()
-    result = conn.execute("SELECT 1").fetchone()[0]
+    result_row = conn.execute("SELECT 1").fetchone()
+    result = result_row[0] if result_row else 0
     return {"status": "ok" if result == 1 else "error"}
 
 
@@ -50,13 +61,167 @@ async def purge_tokens(_: dict = Depends(require_admin)) -> dict:
     conn = get_conn()
     lock = get_write_lock()
     now = datetime.now(timezone.utc)
+    sql_count = "SELECT COUNT(*) FROM refresh_tokens"
+    sql_delete = "DELETE FROM refresh_tokens WHERE revoked = true OR expires_at <= ?"
     async with lock:
-        before = conn.execute(
-            "SELECT COUNT(*) FROM refresh_tokens").fetchone()[0]
-        conn.execute(
-            "DELETE FROM refresh_tokens WHERE revoked = true OR expires_at <= ?", [
-                now]
-        )
-        after = conn.execute(
-            "SELECT COUNT(*) FROM refresh_tokens").fetchone()[0]
+        before_row = conn.execute(sql_count).fetchone()
+        before = before_row[0] if before_row else 0
+
+        conn.execute(sql_delete, [now])
+
+        after_row = conn.execute(sql_count).fetchone()
+        after = after_row[0] if after_row else 0
+
         return {"deleted": before - after, "remaining": after}
+
+
+@router.get("/models/status")
+async def get_model_status(_: dict = Depends(require_admin)) -> dict[str, Any]:
+    """Return model cache status: last update time and model count."""
+    conn = get_conn()
+    return models_service.get_cache_status(conn)
+
+
+@router.get("/models")
+async def get_all_models(_: dict = Depends(require_admin)) -> list[dict[str, Any]]:
+    """Return all cached models."""
+    conn = get_conn()
+    return models_service.get_cached_models(conn)
+
+
+@router.post("/models/refresh", status_code=200)
+async def refresh_models(
+    _: dict = Depends(require_admin),
+) -> dict[str, str | int | None]:
+    """Force a refresh of the model cache from OpenRouter."""
+    conn = get_conn()
+    lock = get_write_lock()
+    async with lock:
+        count = await models_service.refresh_models_cache(conn)
+    status = models_service.get_cache_status(conn)
+    return {
+        "status": "ok",
+        "refreshed": count,
+        "total_models": status.get("model_count"),
+        "last_updated": status.get("last_updated"),
+    }
+
+
+@router.get("/videos")
+async def admin_list_video_jobs(_: dict = Depends(require_admin)) -> list[dict[str, Any]]:
+    """Return all video generation jobs across all users."""
+    conn = get_conn()
+    rows = conn.execute(
+        """
+        SELECT
+            v.id, v.job_id, v.user_id, u.email, v.model_id, v.prompt,
+            v.status, v.video_url, v.created_at, v.updated_at
+        FROM generated_videos v
+        LEFT JOIN users u ON v.user_id = u.id
+        ORDER BY v.created_at DESC
+        """
+    ).fetchall()
+    return [
+        {
+            "id": str(row[0]),
+            "job_id": row[1],
+            "user_id": str(row[2]),
+            "user_email": row[3],
+            "model_id": row[4],
+            "prompt": row[5],
+            "status": row[6],
+            "video_url": row[7],
+            "created_at": row[8].isoformat() if row[8] else None,
+            "updated_at": row[9].isoformat() if row[9] else None,
+        }
+        for row in rows
+    ]
+
+
+@router.post("/videos/{job_id}/cancel", status_code=200)
+async def admin_cancel_video_job(job_id: str, _: dict = Depends(require_admin)) -> dict[str, str]:
+    """Mark a video job as 'cancelled'. Does not stop the provider job."""
+    conn = get_conn()
+    lock = get_write_lock()
+    now = datetime.now(timezone.utc)
+    async with lock:
+        conn.execute(
+            "UPDATE generated_videos SET status = 'cancelled', updated_at = ? WHERE id = ?", [
+                now, job_id]
+        )
+    return {"status": "ok", "job_id": job_id}
+
+
+@router.post("/videos/purge", status_code=200)
+async def admin_purge_video_jobs(_: dict = Depends(require_admin)) -> dict[str, Any]:
+    """Delete all completed, failed, or cancelled jobs older than 30 days."""
+    conn = get_conn()
+    lock = get_write_lock()
+    thirty_days_ago = datetime.now(
+        timezone.utc) - timedelta(days=30)
+
+    sql_count = "SELECT COUNT(*) FROM generated_videos"
+    sql_delete = """
+        DELETE FROM generated_videos
+        WHERE status IN ('completed', 'failed', 'cancelled')
+        AND updated_at < ?
+    """
+
+    async with lock:
+        before_row = conn.execute(sql_count).fetchone()
+        before = before_row[0] if before_row else 0
+
+        conn.execute(sql_delete, [thirty_days_ago])
+
+        after_row = conn.execute(sql_count).fetchone()
+        after = after_row[0] if after_row else 0
+
+    return {"deleted": before - after, "remaining": after}
+
+
+@router.post("/videos/timed-out", status_code=200)
+async def admin_mark_timed_out(_: dict = Depends(require_admin)) -> dict[str, int]:
+    """Mark video jobs that have been in 'queued' or 'processing' status for too long as 'failed'."""
+    conn = get_conn()
+    count = mark_timed_out_video_jobs(conn, timeout_minutes=120)
+    return {"timed_out": count}
+
+
+@router.post("/videos/{job_id}/retry", status_code=200)
+async def admin_retry_video_job(job_id: str, _: dict = Depends(require_admin)) -> dict[str, str]:
+    """Reset a failed or cancelled video job back to 'queued' for reprocessing."""
+    conn = get_conn()
+    lock = get_write_lock()
+    now = datetime.now(timezone.utc)
+    async with lock:
+        row = conn.execute(
+            "SELECT status FROM generated_videos WHERE id = ?", [job_id]
+        ).fetchone()
+        if row is None:
+            from fastapi import HTTPException
+            raise HTTPException(status_code=404, detail="Job not found")
+        if row[0] not in ("failed", "cancelled"):
+            from fastapi import HTTPException
+            raise HTTPException(
+                status_code=400, detail=f"Cannot retry job with status '{row[0]}'")
+        conn.execute(
+            "UPDATE generated_videos SET status = 'queued', updated_at = ? WHERE id = ?",
+            [now, job_id],
+        )
+    return {"status": "ok", "job_id": job_id}
+
+
+@router.delete("/videos/{job_id}", status_code=200)
+async def admin_delete_video_job(job_id: str, _: dict = Depends(require_admin)) -> dict[str, str]:
+    """Permanently delete a video job record."""
+    conn = get_conn()
+    lock = get_write_lock()
+    async with lock:
+        row = conn.execute(
+            "SELECT id FROM generated_videos WHERE id = ?", [job_id]
+        ).fetchone()
+        if row is None:
+            from fastapi import HTTPException
+            raise HTTPException(status_code=404, detail="Job not found")
+        conn.execute("DELETE FROM generated_videos WHERE id = ?", [job_id])
+    return {"status": "ok", "job_id": job_id}

backend/app/routers/auth.py

@@ -24,7 +24,8 @@ async def register(body: RegisterRequest) -> dict:
     try:
         user = await register_user(body.email, body.password)
     except ValueError as exc:
-        raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail=str(exc))
+        raise HTTPException(
+            status_code=status.HTTP_409_CONFLICT, detail=str(exc))
     return {"id": user["id"], "email": user["email"], "role": user["role"]}
 
 
@@ -40,7 +41,8 @@ async def login(body: LoginRequest) -> TokenResponse:
     jti = str(uuid.uuid4())
     await store_refresh_token(user["id"], jti)
     return TokenResponse(
-        access_token=create_access_token(user["id"], user["email"], user["role"]),
+        access_token=create_access_token(
+            user["id"], user["email"], user["role"]),
         refresh_token=create_refresh_token(user["id"], jti),
     )
 
@@ -73,9 +75,8 @@ async def refresh(body: RefreshRequest) -> TokenResponse:
 
     from ..db import get_conn
     conn = get_conn()
-    row = conn.execute(
-        "SELECT email, role FROM users WHERE id = ?", [user_id]
-    ).fetchone()
+    sql_fetch = "SELECT email, role FROM users WHERE id = ?"
+    row = conn.execute(sql_fetch, [user_id]).fetchone()
     if row is None:
         raise credentials_error
 

backend/app/routers/generate.py

@@ -1,6 +1,11 @@
 """Generate router: text, image, video, and image-to-video generation."""
+import json
+from datetime import datetime, timezone
+
+import httpx
 from fastapi import APIRouter, Depends, HTTPException, status
 
+from ..db import get_conn, get_write_lock
 from ..dependencies import get_current_user
 from ..models.ai import (
     ImageRequest,
@@ -13,6 +18,7 @@ from ..models.ai import (
     VideoResponse,
 )
 from ..services import openrouter
+from ..services.models import get_model_output_modalities
 
 router = APIRouter(prefix="/generate", tags=["generate"])
 
@@ -23,6 +29,13 @@ async def generate_text(
     _: dict = Depends(get_current_user),
 ) -> TextResponse:
     """Generate text from a prompt using a chat model."""
+    if body.messages:
+        messages = [{"role": m.role, "content": m.content}
+                    for m in body.messages]
+        if body.system_prompt and (not messages or messages[0]["role"] != "system"):
+            messages.insert(
+                0, {"role": "system", "content": body.system_prompt})
+    else:
         messages = []
         if body.system_prompt:
             messages.append({"role": "system", "content": body.system_prompt})
@@ -55,160 +68,336 @@ async def generate_text(
 @router.post("/image", response_model=ImageResponse)
 async def generate_image(
     body: ImageRequest,
-    _: dict = Depends(get_current_user),
+    current_user: dict = Depends(get_current_user),
 ) -> ImageResponse:
-    """Generate images from a text prompt."""
-    # Detect if model uses chat completions (FLUX, GPT-5 Image Mini) vs /images/generations (DALL-E)
-    chat_models = {"black-forest-labs/flux.2-klein-4b",
-                   "openai/gpt-5-image-mini"}
-    is_chat_model = body.model.lower() in {m.lower() for m in chat_models} or \
-        any(m in body.model.lower() for m in ["flux", "gpt-5-image-mini"])
+    """Generate images from a prompt using the chat completions endpoint.
 
+    All OpenRouter image models use /chat/completions with a modalities param.
+    Models that output only images use ["image"]; those that also output text
+    use ["image", "text"]. We look this up from the model cache; default to
+    ["image", "text"] when the model is not yet cached.
+    """
+    # Determine modalities from cache; default ["image", "text"] works for most models
     try:
-        if is_chat_model:
-            image_config = {}
+        conn = get_conn()
+        cached_modalities = get_model_output_modalities(conn, body.model)
+    except Exception:
+        cached_modalities = []
+
+    if cached_modalities:
+        # If cache says model only outputs image (no text), use ["image"]
+        modalities = ["image"] if set(cached_modalities) == {
+            "image"} else ["image", "text"]
+    else:
+        # Safe default: ["image", "text"]; works for Gemini, GPT-image etc.
+        # For image-only models that fail with this, the error surfaces to the user.
+        modalities = ["image", "text"]
+
+    image_config: dict = {}
     if body.aspect_ratio:
         image_config["aspect_ratio"] = body.aspect_ratio
     if body.image_size:
         image_config["image_size"] = body.image_size
+
+    try:
         result = await openrouter.generate_image_chat(
             model=body.model,
             prompt=body.prompt,
-                modalities=[
-                    "image", "text"] if "gpt-5-image-mini" in body.model.lower() else ["image"],
+            modalities=modalities,
             image_config=image_config if image_config else None,
         )
-        else:
-            result = await openrouter.generate_image(
-                model=body.model,
-                prompt=body.prompt,
-                n=body.n,
-                size=body.size,
-            )
     except Exception as exc:
         raise HTTPException(
             status_code=status.HTTP_502_BAD_GATEWAY, detail=f"OpenRouter error: {exc}")
 
     try:
-        if is_chat_model:
-            # Chat completions response: choices[0].message.images[].image_url.url
-            images = []
         message = result.get("choices", [{}])[0].get("message", {})
+        images = []
         for item in message.get("images", []):
             img_url = item.get("image_url", {}).get("url")
             images.append(ImageResult(
                 url=img_url,
                 b64_json=None,
-                    revised_prompt=message.get("content"),
+                revised_prompt=message.get("content") or None,
             ))
-            return ImageResponse(
-                id=result.get("id", ""),
-                model=result.get("model", body.model),
-                images=images,
+        if not images:
+            raise HTTPException(
+                status_code=status.HTTP_502_BAD_GATEWAY,
+                detail="No images returned by model. Verify the model supports image generation.",
             )
+
+        # Persist each image to DB
+        user_id = current_user.get("id") or current_user.get("sub")
+        now = datetime.now(timezone.utc).replace(tzinfo=None)
+        stored: list[ImageResult] = []
+        sql_insert = "INSERT INTO generated_images (user_id, model_id, prompt, image_data, created_at) VALUES (?, ?, ?, ?, ?) RETURNING id"
+        async with get_write_lock():
+            conn = get_conn()
+            for img in images:
+                if img.url:
+                    row = conn.execute(
+                        sql_insert, [user_id, body.model, body.prompt, img.url, now],).fetchone()
+                    image_id = str(row[0]) if row else None
                 else:
-            # /images/generations response: data[].url
-            images = [
-                ImageResult(
-                    url=item.get("url"),
-                    b64_json=item.get("b64_json"),
-                    revised_prompt=item.get("revised_prompt"),
-                )
-                for item in result.get("data", [])
-            ]
+                    image_id = None
+                stored.append(ImageResult(
+                    url=img.url,
+                    b64_json=img.b64_json,
+                    revised_prompt=img.revised_prompt,
+                    image_id=image_id,
+                ))
+
         return ImageResponse(
             id=result.get("id", ""),
             model=result.get("model", body.model),
-                images=images,
+            images=stored,
         )
+    except HTTPException:
+        raise
     except (KeyError, TypeError) as exc:
         raise HTTPException(status_code=status.HTTP_502_BAD_GATEWAY,
                             detail=f"Unexpected response format: {exc}")
 
 
+@router.get("/images")
+async def list_generated_images(
+    current_user: dict = Depends(get_current_user),
+) -> list[dict]:
+    """Return all generated images for the current user, newest first."""
+    user_id = current_user.get("id") or current_user.get("sub")
+    conn = get_conn()
+    sql_fetch = "SELECT id, model_id, prompt, image_data, created_at FROM generated_images WHERE user_id = ? ORDER BY created_at DESC"
+    rows = conn.execute(sql_fetch, [user_id]).fetchall()
+    return [
+        {
+            "id": str(r[0]),
+            "model_id": r[1],
+            "prompt": r[2],
+            "image_data": r[3],
+            "created_at": r[4].isoformat() if r[4] else None,
+        }
+        for r in rows
+    ]
+
+
+@router.get("/images/{image_id}")
+async def get_generated_image(
+    image_id: str,
+    current_user: dict = Depends(get_current_user),
+) -> dict:
+    """Return details for a single generated image."""
+    user_id = current_user.get("id") or current_user.get("sub")
+    conn = get_conn()
+    row = conn.execute(
+        """SELECT id, model_id, prompt, image_data, created_at
+           FROM generated_images
+           WHERE id = ? AND user_id = ?""",
+        [image_id, user_id],
+    ).fetchone()
+    if not row:
+        raise HTTPException(status_code=404, detail="Image not found")
+    return {
+        "id": str(row[0]),
+        "model_id": row[1],
+        "prompt": row[2],
+        "image_data": row[3],
+        "created_at": row[4].isoformat() if row[4] else None,
+    }
+
+
 @router.post("/video", response_model=VideoResponse)
 async def generate_video(
     body: VideoRequest,
-    _: dict = Depends(get_current_user),
+    current_user: dict = Depends(get_current_user),
 ) -> VideoResponse:
-    """Generate a video from a text prompt."""
-    try:
-        result = await openrouter.generate_video(
-            model=body.model,
-            prompt=body.prompt,
-            duration_seconds=body.duration_seconds,
-            aspect_ratio=body.aspect_ratio,
-            resolution=body.resolution,
-        )
-    except Exception as exc:
-        raise HTTPException(
-            status_code=status.HTTP_502_BAD_GATEWAY, detail=f"OpenRouter error: {exc}")
-
-    urls = result.get("unsigned_urls") or result.get("video_urls")
+    """Queue a text-to-video generation job for background processing."""
+    user_id = current_user.get("id") or current_user.get("sub")
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+    request_params = json.dumps({
+        "model": body.model,
+        "prompt": body.prompt,
+        "duration_seconds": body.duration_seconds,
+        "aspect_ratio": body.aspect_ratio,
+        "resolution": body.resolution,
+    })
+    db_id = None
+    async with get_write_lock():
+        conn = get_conn()
+        row = conn.execute(
+            """INSERT INTO generated_videos
+               (user_id, job_id, model_id, prompt, status, request_params, generation_type, created_at, updated_at)
+               VALUES (?, ?, ?, ?, 'queued', ?, 'text_to_video', ?, ?) RETURNING id""",
+            [user_id, "", body.model, body.prompt, request_params, now, now],
+        ).fetchone()
+        if row:
+            db_id = str(row[0])
     return VideoResponse(
-        id=result.get("id", ""),
+        id="",
+        db_id=db_id,
         model=body.model,
-        status=result.get("status", "queued"),
-        polling_url=result.get("polling_url"),
-        video_urls=urls,
-        video_url=(urls or [None])[0],
-        error=result.get("error"),
-        metadata=result.get("metadata"),
+        status="queued",
     )
 
 
 @router.post("/video/from-image", response_model=VideoResponse)
 async def generate_video_from_image(
     body: VideoFromImageRequest,
-    _: dict = Depends(get_current_user),
+    current_user: dict = Depends(get_current_user),
 ) -> VideoResponse:
-    """Generate a video from an image and a text prompt."""
-    try:
-        result = await openrouter.generate_video_from_image(
-            model=body.model,
-            image_url=body.image_url,
-            prompt=body.prompt,
-            duration_seconds=body.duration_seconds,
-            aspect_ratio=body.aspect_ratio,
-            resolution=body.resolution,
-        )
-    except Exception as exc:
-        raise HTTPException(
-            status_code=status.HTTP_502_BAD_GATEWAY, detail=f"OpenRouter error: {exc}")
-
-    urls = result.get("unsigned_urls") or result.get("video_urls")
+    """Queue an image-to-video generation job for background processing."""
+    user_id = current_user.get("id") or current_user.get("sub")
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+    request_params = json.dumps({
+        "model": body.model,
+        "image_url": body.image_url,
+        "prompt": body.prompt,
+        "duration_seconds": body.duration_seconds,
+        "aspect_ratio": body.aspect_ratio,
+        "resolution": body.resolution,
+    })
+    db_id = None
+    async with get_write_lock():
+        conn = get_conn()
+        row = conn.execute(
+            """INSERT INTO generated_videos
+               (user_id, job_id, model_id, prompt, status, request_params, generation_type, created_at, updated_at)
+               VALUES (?, ?, ?, ?, 'queued', ?, 'image_to_video', ?, ?) RETURNING id""",
+            [user_id, "", body.model, body.prompt, request_params, now, now],
+        ).fetchone()
+        if row:
+            db_id = str(row[0])
     return VideoResponse(
-        id=result.get("id", ""),
+        id="",
+        db_id=db_id,
         model=body.model,
-        status=result.get("status", "queued"),
-        polling_url=result.get("polling_url"),
-        video_urls=urls,
-        video_url=(urls or [None])[0],
-        error=result.get("error"),
-        metadata=result.get("metadata"),
+        status="queued",
     )
 
 
 @router.get("/video/status", response_model=VideoResponse)
 async def poll_video_status(
     polling_url: str,
-    _: dict = Depends(get_current_user),
+    current_user: dict = Depends(get_current_user),
 ) -> VideoResponse:
-    """Poll the status of a video generation job via its polling_url."""
+    """Poll status of a video generation job; updates DB row when completed/failed."""
     try:
         result = await openrouter.poll_video_status(polling_url)
     except Exception as exc:
         raise HTTPException(
             status_code=status.HTTP_502_BAD_GATEWAY, detail=f"OpenRouter error: {exc}")
 
+    job_status = result.get("status", "processing")
     urls = result.get("unsigned_urls") or result.get("video_urls")
+    video_url = (urls or [None])[0]
+
+    # Update DB row for this job when terminal state reached
+    if job_status in ("completed", "failed"):
+        now = datetime.now(timezone.utc).replace(tzinfo=None)
+        async with get_write_lock():
+            conn = get_conn()
+            conn.execute(
+                """UPDATE generated_videos
+                SET status = ?, video_url = ?, updated_at = ?
+                WHERE job_id = ?""",
+                [job_status, video_url, now, result.get("id", "")],
+            )
+
     return VideoResponse(
         id=result.get("id", ""),
         model=result.get("model", ""),
-        status=result.get("status", "processing"),
+        status=job_status,
         polling_url=result.get("polling_url"),
         video_urls=urls,
-        video_url=(urls or [None])[0],
+        video_url=video_url,
         error=result.get("error"),
         metadata=result.get("metadata"),
     )
+
+
+@router.get("/videos")
+async def list_generated_videos(
+    current_user: dict = Depends(get_current_user),
+) -> list[dict]:
+    """Return all generated video jobs for the current user, newest first."""
+    user_id = current_user.get("id") or current_user.get("sub")
+    conn = get_conn()
+    rows = conn.execute(
+        """SELECT id, job_id, model_id, prompt, polling_url, status, video_url, error, created_at
+           FROM generated_videos
+           WHERE user_id = ?
+           ORDER BY created_at DESC""",
+        [user_id],
+    ).fetchall()
+    return [
+        {
+            "id": str(r[0]),
+            "job_id": r[1],
+            "model_id": r[2],
+            "prompt": r[3],
+            "polling_url": r[4],
+            "status": r[5],
+            "video_url": r[6],
+            "error": r[7],
+            "created_at": r[8].isoformat() if r[8] else None,
+        }
+        for r in rows
+    ]
+
+
+@router.get("/videos/{video_id}")
+async def get_generated_video(
+    video_id: str,
+    current_user: dict = Depends(get_current_user),
+) -> dict:
+    """Return details for a single video generation job."""
+    user_id = current_user.get("id") or current_user.get("sub")
+    conn = get_conn()
+    row = conn.execute(
+        """SELECT id, job_id, model_id, prompt, polling_url, status, video_url, error, created_at, updated_at
+           FROM generated_videos
+           WHERE id = ? AND user_id = ?""",
+        [video_id, user_id],
+    ).fetchone()
+    if not row:
+        raise HTTPException(status_code=404, detail="Video job not found")
+    return {
+        "id": str(row[0]),
+        "job_id": row[1],
+        "model_id": row[2],
+        "prompt": row[3],
+        "polling_url": row[4],
+        "status": row[5],
+        "video_url": row[6],
+        "error": row[7],
+        "created_at": row[8].isoformat() if row[8] else None,
+        "updated_at": row[9].isoformat() if row[9] else None,
+    }
+
+
+@router.post("/videos/{video_id}/cancel", status_code=200)
+async def cancel_video_job(
+    video_id: str,
+    current_user: dict = Depends(get_current_user),
+) -> dict[str, str]:
+    """Mark a video job as 'cancelled' if it belongs to the current user and is not terminal."""
+    user_id = current_user.get("id") or current_user.get("sub")
+    conn = get_conn()
+    row = conn.execute(
+        "SELECT status FROM generated_videos WHERE id = ? AND user_id = ?",
+        [video_id, user_id],
+    ).fetchone()
+    if not row:
+        raise HTTPException(status_code=404, detail="Video job not found")
+    job_status = row[0]
+    if job_status in ("completed", "failed", "cancelled"):
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"Cannot cancel job with status '{job_status}'",
+        )
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+    async with get_write_lock():
+        conn.execute(
+            "UPDATE generated_videos SET status = 'cancelled', updated_at = ? WHERE id = ?",
+            [now, video_id],
+        )
+    return {"status": "ok", "job_id": video_id}

backend/app/routers/images.py

@@ -0,0 +1,150 @@
+"""Images router: upload reference images and list user's uploads."""
+import os
+import uuid
+
+from fastapi import APIRouter, Depends, HTTPException, UploadFile, status
+from fastapi.responses import FileResponse
+
+from ..db import get_conn, get_write_lock
+from ..dependencies import get_current_user
+
+router = APIRouter(prefix="/images", tags=["images"])
+
+UPLOAD_DIR = os.getenv("UPLOAD_DIR", "data/uploads")
+MAX_SIZE_BYTES = 10 * 1024 * 1024  # 10 MB
+ALLOWED_CONTENT_TYPES = {"image/jpeg", "image/png", "image/webp", "image/gif"}
+
+
+@router.post("/upload", status_code=status.HTTP_201_CREATED)
+async def upload_image(
+    file: UploadFile,
+    current_user: dict = Depends(get_current_user),
+) -> dict:
+    """Upload a reference image and store metadata in DuckDB."""
+    if file.content_type not in ALLOWED_CONTENT_TYPES:
+        raise HTTPException(
+            status_code=status.HTTP_415_UNSUPPORTED_MEDIA_TYPE,
+            detail=f"Unsupported content type '{file.content_type}'. Allowed: {sorted(ALLOWED_CONTENT_TYPES)}",
+        )
+
+    data = await file.read()
+    if len(data) > MAX_SIZE_BYTES:
+        raise HTTPException(
+            status_code=status.HTTP_413_REQUEST_ENTITY_TOO_LARGE,
+            detail=f"File exceeds maximum allowed size of {MAX_SIZE_BYTES // (1024*1024)} MB.",
+        )
+
+    user_id = current_user["id"]
+    image_id = str(uuid.uuid4())
+    ext = (file.filename or "").rsplit(
+        ".", 1)[-1].lower() if "." in (file.filename or "") else "bin"
+    safe_filename = f"{image_id}.{ext}"
+    user_dir = os.path.join(UPLOAD_DIR, user_id)
+    os.makedirs(user_dir, exist_ok=True)
+    file_path = os.path.join(user_dir, safe_filename)
+
+    with open(file_path, "wb") as f:
+        f.write(data)
+
+    async with get_write_lock():
+        conn = get_conn()
+        conn.execute(
+            """
+            INSERT INTO uploaded_images (id, user_id, filename, content_type, file_path, size_bytes)
+            VALUES (?, ?, ?, ?, ?, ?)
+            """,
+            [image_id, user_id, file.filename or safe_filename,
+                file.content_type, file_path, len(data)],
+        )
+
+    return {
+        "id": image_id,
+        "filename": file.filename or safe_filename,
+        "content_type": file.content_type,
+        "size_bytes": len(data),
+    }
+
+
+@router.get("/", status_code=status.HTTP_200_OK)
+async def list_images(
+    current_user: dict = Depends(get_current_user),
+) -> list[dict]:
+    """Return all uploaded images for the current user."""
+    conn = get_conn()
+    rows = conn.execute(
+        """
+        SELECT id, filename, content_type, size_bytes, created_at
+        FROM uploaded_images
+        WHERE user_id = ?
+        ORDER BY created_at DESC
+        """,
+        [current_user["id"]],
+    ).fetchall()
+
+    return [
+        {
+            "id": str(row[0]),
+            "filename": row[1],
+            "content_type": row[2],
+            "size_bytes": row[3],
+            "created_at": row[4].isoformat() if row[4] else None,
+        }
+        for row in rows
+    ]
+
+
+@router.get("/{image_id}", status_code=status.HTTP_200_OK)
+async def get_image_details(
+    image_id: str,
+    current_user: dict = Depends(get_current_user),
+) -> dict:
+    """Return metadata for a single uploaded image."""
+    conn = get_conn()
+    row = conn.execute(
+        """
+        SELECT id, filename, content_type, size_bytes, created_at
+        FROM uploaded_images
+        WHERE id = ? AND user_id = ?
+        """,
+        [image_id, current_user["id"]],
+    ).fetchone()
+
+    if not row:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND, detail="Image not found"
+        )
+
+    return {
+        "id": str(row[0]),
+        "filename": row[1],
+        "content_type": row[2],
+        "size_bytes": row[3],
+        "created_at": row[4].isoformat() if row[4] else None,
+    }
+
+
+@router.get("/{image_id}/file", status_code=status.HTTP_200_OK)
+async def serve_image(
+    image_id: str,
+    current_user: dict = Depends(get_current_user),
+) -> FileResponse:
+    """Serve the raw image file. Only accessible by the owning user."""
+    conn = get_conn()
+    row = conn.execute(
+        "SELECT file_path, content_type, user_id FROM uploaded_images WHERE id = ?",
+        [image_id],
+    ).fetchone()
+
+    if row is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND, detail="Image not found.")
+    if str(row[2]) != current_user["id"]:
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN, detail="Access denied.")
+
+    file_path: str = row[0]
+    if not os.path.isfile(file_path):
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND, detail="Image file missing.")
+
+    return FileResponse(file_path, media_type=row[1])

backend/app/routers/models.py

@@ -0,0 +1,47 @@
+"""Models router: list and refresh the OpenRouter model cache."""
+from fastapi import APIRouter, Depends, HTTPException, Query, status
+
+from ..db import get_conn, get_write_lock
+from ..dependencies import get_current_user, require_admin
+from ..services import models as models_service
+
+router = APIRouter(prefix="/models", tags=["models"])
+
+
+@router.get("/")
+async def list_models(
+    modality: str | None = Query(
+        None,
+        description="Filter by output modality: text, image, video, audio",
+    ),
+    _: dict = Depends(get_current_user),
+):
+    """Return cached models. Auto-refreshes cache if stale (older than 24 h)."""
+    conn = get_conn()
+    if models_service.is_cache_stale(conn):
+        async with get_write_lock():
+            # Re-check inside lock to avoid redundant parallel refreshes
+            if models_service.is_cache_stale(conn):
+                try:
+                    await models_service.refresh_models_cache(conn)
+                except Exception as exc:
+                    raise HTTPException(
+                        status_code=status.HTTP_502_BAD_GATEWAY,
+                        detail=f"Failed to refresh model cache: {exc}",
+                    )
+    return models_service.get_cached_models(conn, modality)
+
+
+@router.post("/refresh", status_code=200)
+async def refresh_models(_: dict = Depends(require_admin)):
+    """Force-refresh the model cache from OpenRouter. Admin only."""
+    conn = get_conn()
+    async with get_write_lock():
+        try:
+            count = await models_service.refresh_models_cache(conn)
+        except Exception as exc:
+            raise HTTPException(
+                status_code=status.HTTP_502_BAD_GATEWAY,
+                detail=f"OpenRouter error: {exc}",
+            )
+    return {"refreshed": count}

backend/app/services/auth.py

@@ -35,7 +35,8 @@ def verify_password(plain: str, hashed: str) -> bool:
 # --- Tokens ---
 
 def create_access_token(user_id: str, email: str, role: str) -> str:
-    expire = datetime.now(timezone.utc) + timedelta(minutes=ACCESS_TOKEN_EXPIRE_MINUTES)
+    expire = datetime.now(timezone.utc) + \
+        timedelta(minutes=ACCESS_TOKEN_EXPIRE_MINUTES)
     payload = {
         "sub": user_id,
         "email": email,
@@ -47,7 +48,8 @@ def create_access_token(user_id: str, email: str, role: str) -> str:
 
 
 def create_refresh_token(user_id: str, jti: str) -> str:
-    expire = datetime.now(timezone.utc) + timedelta(days=REFRESH_TOKEN_EXPIRE_DAYS)
+    expire = datetime.now(timezone.utc) + \
+        timedelta(days=REFRESH_TOKEN_EXPIRE_DAYS)
     payload = {
         "sub": user_id,
         "jti": jti,
@@ -68,28 +70,25 @@ async def register_user(email: str, password: str) -> dict[str, Any]:
     """Insert a new user. Returns the created user row."""
     conn = get_conn()
     lock = get_write_lock()
+    sql_check = "SELECT id FROM users WHERE email = ?"
+    sql_insert = "INSERT INTO users (email, password_hash) VALUES (?, ?)"
+    sql_fetch = "SELECT id, email, role FROM users WHERE email = ?"
     async with lock:
-        existing = conn.execute(
-            "SELECT id FROM users WHERE email = ?", [email]
-        ).fetchone()
+        existing = conn.execute(sql_check, [email]).fetchone()
         if existing:
             raise ValueError("Email already registered.")
-        conn.execute(
-            "INSERT INTO users (email, password_hash) VALUES (?, ?)",
-            [email, hash_password(password)],
-        )
-        row = conn.execute(
-            "SELECT id, email, role FROM users WHERE email = ?", [email]
-        ).fetchone()
+        conn.execute(sql_insert, [email, hash_password(password)],)
+        row = conn.execute(sql_fetch, [email]).fetchone()
+    if row is None:
+        raise RuntimeError("Failed to fetch user after registration.")
     return {"id": str(row[0]), "email": row[1], "role": row[2]}
 
 
 async def authenticate_user(email: str, password: str) -> dict[str, Any] | None:
     """Return user dict if credentials are valid, else None."""
     conn = get_conn()
-    row = conn.execute(
-        "SELECT id, email, password_hash, role FROM users WHERE email = ?", [email]
-    ).fetchone()
+    sql_fetch = "SELECT id, email, password_hash, role FROM users WHERE email = ?"
+    row = conn.execute(sql_fetch, [email]).fetchone()
     if row is None or not verify_password(password, row[2]):
         return None
     return {"id": str(row[0]), "email": row[1], "role": row[3]}
@@ -99,34 +98,30 @@ async def store_refresh_token(user_id: str, jti: str) -> None:
     """Persist a refresh token JTI in the database."""
     conn = get_conn()
     lock = get_write_lock()
+    sql_insert = "INSERT INTO refresh_tokens (jti, user_id, expires_at) VALUES (?, ?, ?)"
     from datetime import timedelta
-    expires_at = datetime.now(timezone.utc) + timedelta(days=REFRESH_TOKEN_EXPIRE_DAYS)
+    expires_at = datetime.now(timezone.utc) + \
+        timedelta(days=REFRESH_TOKEN_EXPIRE_DAYS)
     async with lock:
-        conn.execute(
-            "INSERT INTO refresh_tokens (jti, user_id, expires_at) VALUES (?, ?, ?)",
-            [jti, user_id, expires_at],
-        )
+        conn.execute(sql_insert, [jti, user_id, expires_at])
 
 
 async def revoke_refresh_token(jti: str) -> None:
     """Mark a refresh token as revoked."""
     conn = get_conn()
     lock = get_write_lock()
+    sql_update = "UPDATE refresh_tokens SET revoked = true WHERE jti = ?"
     async with lock:
-        conn.execute(
-            "UPDATE refresh_tokens SET revoked = true WHERE jti = ?", [jti]
-        )
+        conn.execute(sql_update, [jti])
 
 
 async def validate_refresh_token_jti(jti: str, user_id: str) -> bool:
     """Return True if the JTI exists, is not revoked, and belongs to user_id."""
     conn = get_conn()
     now = datetime.now(timezone.utc)
-    row = conn.execute(
-        """
+    sql_select = """
         SELECT 1 FROM refresh_tokens
         WHERE jti = ? AND user_id = ? AND revoked = false AND expires_at > ?
-        """,
-        [jti, user_id, now],
-    ).fetchone()
+        """
+    row = conn.execute(sql_select, [jti, user_id, now]).fetchone()
     return row is not None

backend/app/services/models.py

@@ -0,0 +1,246 @@
+"""Model cache service: fetch from OpenRouter, store in DuckDB."""
+import json
+from datetime import datetime, timedelta, timezone
+from typing import Any
+
+import duckdb
+
+from . import openrouter
+
+CACHE_TTL_HOURS = 24
+
+
+def _normalize_modality(raw: str) -> str:
+    """Normalize OpenRouter modality labels to canonical values."""
+    value = (raw or "").strip().lower()
+    if value in {"text", "image", "video", "audio", "embeddings", "embedding"}:
+        return "embeddings" if value == "embedding" else value
+    if "image" in value:
+        return "image"
+    if "video" in value:
+        return "video"
+    if "audio" in value:
+        return "audio"
+    if "embed" in value:
+        return "embeddings"
+    return "text"
+
+
+def _parse_modality(raw_modality: str) -> str:
+    """Extract output modality from OpenRouter architecture.modality string.
+
+    Examples: "text->text", "text+image->text", "text->image", "text->video"
+    """
+    output = raw_modality.split(
+        "->", 1)[-1] if "->" in raw_modality else raw_modality
+    return _normalize_modality(output)
+
+
+def _extract_output_modality(model: dict[str, Any]) -> str:
+    """Extract output modality using OpenRouter schema, fallback to legacy field."""
+    architecture = model.get("architecture") or {}
+
+    output_modalities = architecture.get(
+        "output_modalities") or model.get("output_modalities")
+    if isinstance(output_modalities, list) and output_modalities:
+        return _normalize_modality(str(output_modalities[0]))
+
+    raw_modality = architecture.get(
+        "modality") or model.get("modality") or "text->text"
+    if isinstance(raw_modality, str):
+        return _parse_modality(raw_modality)
+    return "text"
+
+
+async def _fetch_models_for_cache() -> list[dict[str, Any]]:
+    """Fetch broad + modality-specific lists and merge unique models by id."""
+    by_id: dict[str, dict[str, Any]] = {}
+
+    # Primary fetch: all modalities (per OpenRouter docs).
+    primary = await openrouter.list_models(output_modalities="all")
+    for model in primary:
+        model_id = model.get("id")
+        if model_id:
+            by_id[model_id] = model
+
+    # Warmup fetches: some providers surface better results with explicit modality filter.
+    for modality in ("image", "video", "audio", "embeddings", "text"):
+        try:
+            subset = await openrouter.list_models(output_modalities=modality)
+        except Exception:
+            continue
+        for model in subset:
+            model_id = model.get("id")
+            if model_id and model_id not in by_id:
+                by_id[model_id] = model
+
+    return list(by_id.values())
+
+
+async def refresh_models_cache(conn: duckdb.DuckDBPyConnection) -> int:
+    """Fetch all models from OpenRouter and replace the cache. Returns count stored."""
+    raw = await _fetch_models_for_cache()
+    # Use naive UTC to avoid DuckDB TIMESTAMP tz-stripping inconsistencies
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+
+    conn.execute("DELETE FROM models_cache")
+    count = 0
+    for m in raw:
+        modality = _extract_output_modality(m)
+        pricing = m.get("pricing")
+        model_id = m.get("id", "")
+        if not model_id:
+            continue
+        # Full output_modalities array from architecture (for proper modalities param in image gen)
+        architecture = m.get("architecture") or {}
+        raw_output_modalities: list | None = (
+            architecture.get("output_modalities") or m.get("output_modalities")
+        )
+        output_modalities_json: str | None = (
+            json.dumps([_normalize_modality(str(v))
+                       for v in raw_output_modalities])
+            if isinstance(raw_output_modalities, list)
+            else None
+        )
+        conn.execute(
+            """
+            INSERT INTO models_cache (model_id, name, modality, context_length, pricing, fetched_at, output_modalities)
+            VALUES (?, ?, ?, ?, ?, ?, ?)
+            ON CONFLICT (model_id) DO UPDATE SET
+                name = excluded.name,
+                modality = excluded.modality,
+                context_length = excluded.context_length,
+                pricing = excluded.pricing,
+                fetched_at = excluded.fetched_at,
+                output_modalities = excluded.output_modalities
+            """,
+            [
+                model_id,
+                m.get("name", model_id),
+                modality,
+                m.get("context_length"),
+                json.dumps(pricing) if pricing else None,
+                now,
+                output_modalities_json,
+            ],
+        )
+        count += 1
+    return count
+
+
+def is_cache_stale(conn: duckdb.DuckDBPyConnection) -> bool:
+    """Return True if cache is empty or last fetched more than CACHE_TTL_HOURS ago."""
+    row = conn.execute("SELECT MAX(fetched_at) FROM models_cache").fetchone()
+    if not row or row[0] is None:
+        return True
+    last_fetched = row[0]
+    # DuckDB TIMESTAMP is always naive; compare against naive UTC
+    if last_fetched.tzinfo is not None:
+        last_fetched = last_fetched.replace(tzinfo=None)
+    now_naive = datetime.now(timezone.utc).replace(tzinfo=None)
+    return now_naive - last_fetched > timedelta(hours=CACHE_TTL_HOURS)
+
+
+def get_cached_models(
+    conn: duckdb.DuckDBPyConnection,
+    modality: str | None = None,
+) -> list[dict[str, Any]]:
+    """Return cached models, optionally filtered by modality, ordered by name."""
+    if modality:
+        rows = conn.execute(
+            """
+            SELECT model_id, name, modality, context_length, pricing
+            FROM models_cache
+            WHERE modality = ?
+            ORDER BY name
+            """,
+            [modality],
+        ).fetchall()
+    else:
+        rows = conn.execute(
+            """
+            SELECT model_id, name, modality, context_length, pricing
+            FROM models_cache
+            ORDER BY name
+            """
+        ).fetchall()
+
+    result = []
+    for row in rows:
+        pricing = None
+        if row[4]:
+            try:
+                pricing = json.loads(row[4])
+            except (json.JSONDecodeError, TypeError):
+                pricing = None
+        result.append({
+            "id": row[0],
+            "name": row[1],
+            "modality": row[2],
+            "context_length": row[3],
+            "pricing": pricing,
+        })
+    return result
+
+
+def get_model_output_modalities(
+    conn: duckdb.DuckDBPyConnection,
+    model_id: str,
+) -> list[str]:
+    """Return output_modalities list for a model; empty list if not found."""
+    row = conn.execute(
+        "SELECT output_modalities FROM models_cache WHERE model_id = ?",
+        [model_id],
+    ).fetchone()
+    if not row or not row[0]:
+        return []
+    try:
+        return json.loads(row[0])
+    except (json.JSONDecodeError, TypeError):
+        return []
+
+
+def get_cache_status(conn: duckdb.DuckDBPyConnection) -> dict[str, Any]:
+    """Return cache last update time and model count."""
+    row = conn.execute(
+        "SELECT MAX(fetched_at), COUNT(*) FROM models_cache"
+    ).fetchone()
+    last_updated, model_count = (row[0], row[1]) if row else (None, 0)
+    return {"last_updated": last_updated, "model_count": model_count}
+
+
+def mark_timed_out_video_jobs(conn: duckdb.DuckDBPyConnection, timeout_minutes: int = 120) -> int:
+    """Mark video jobs that have been in 'queued' or 'processing' status for too long as 'failed'.
+
+    Returns the number of jobs marked as timed out.
+    """
+    timeout_threshold = datetime.now(
+        timezone.utc) - timedelta(minutes=timeout_minutes)
+
+    # Find timed out jobs
+    timed_out_rows = conn.execute(
+        """
+        SELECT id FROM generated_videos
+        WHERE status IN ('queued', 'processing')
+        AND updated_at < ?
+        """,
+        [timeout_threshold]
+    ).fetchall()
+
+    if not timed_out_rows:
+        return 0
+
+    job_ids = [row[0] for row in timed_out_rows]
+    placeholders = ",".join(["?"] * len(job_ids))
+
+    # Update them to failed
+    conn.execute(
+        f"""
+        UPDATE generated_videos
+        SET status = 'failed', updated_at = ?
+        WHERE id IN ({placeholders})
+        """,
+        [datetime.now(timezone.utc)] + job_ids
+    )
+
+    return len(job_ids)

backend/app/services/openrouter.py

@@ -20,16 +20,29 @@ def _headers() -> dict[str, str]:
         "Authorization": f"Bearer {_api_key()}",
         "Content-Type": "application/json",
         "HTTP-Referer": os.getenv("APP_URL", "https://ai.allucanget.biz"),
-        "X-Title": os.getenv("APP_NAME", "AI Allucanget"),
+        "X-Title": os.getenv("APP_NAME", "All You Can GET AI"),
     }
 
 
-async def list_models() -> list[dict[str, Any]]:
-    """Return available models from OpenRouter."""
+async def list_models(
+    output_modalities: str = "all",
+    category: str | None = None,
+    supported_parameters: str | None = None,
+) -> list[dict[str, Any]]:
+    """Return available models from OpenRouter.
+
+    Docs: GET /models supports query filters like output_modalities.
+    """
     base_url = os.getenv("OPENROUTER_BASE_URL", OPENROUTER_BASE_URL)
+    params: dict[str, str] = {"output_modalities": output_modalities}
+    if category:
+        params["category"] = category
+    if supported_parameters:
+        params["supported_parameters"] = supported_parameters
+
     async with httpx.AsyncClient(timeout=15) as client:
         resp = client.build_request(
-            "GET", f"{base_url}/models", headers=_headers())
+            "GET", f"{base_url}/models", headers=_headers(), params=params)
         response = await client.send(resp)
         response.raise_for_status()
         return response.json().get("data", [])
@@ -82,8 +95,9 @@ async def generate_video(
     duration_seconds: int | None = None,
     aspect_ratio: str = "16:9",
     resolution: str | None = None,
+    generate_audio: bool | None = None,
 ) -> dict[str, Any]:
-    """Request text-to-video generation via OpenRouter."""
+    """Request text-to-video generation via OpenRouter POST /videos."""
     base_url = os.getenv("OPENROUTER_BASE_URL", OPENROUTER_BASE_URL)
     payload: dict[str, Any] = {
         "model": model,
@@ -91,9 +105,12 @@ async def generate_video(
         "aspect_ratio": aspect_ratio,
     }
     if duration_seconds is not None:
-        payload["duration_seconds"] = duration_seconds
+        # API uses 'duration' not 'duration_seconds'
+        payload["duration"] = duration_seconds
     if resolution is not None:
         payload["resolution"] = resolution
+    if generate_audio is not None:
+        payload["generate_audio"] = generate_audio
     async with httpx.AsyncClient(timeout=120) as client:
         resp = client.build_request(
             "POST", f"{base_url}/videos", headers=_headers(), json=payload
@@ -110,19 +127,31 @@ async def generate_video_from_image(
     duration_seconds: int | None = None,
     aspect_ratio: str = "16:9",
     resolution: str | None = None,
+    generate_audio: bool | None = None,
 ) -> dict[str, Any]:
-    """Request image-to-video generation via OpenRouter."""
+    """Request image-to-video generation via OpenRouter POST /videos.
+
+    Uses frame_images array with first_frame as per OpenRouter API spec.
+    """
     base_url = os.getenv("OPENROUTER_BASE_URL", OPENROUTER_BASE_URL)
     payload: dict[str, Any] = {
         "model": model,
-        "image_url": image_url,
         "prompt": prompt,
         "aspect_ratio": aspect_ratio,
+        "frame_images": [
+            {
+                "type": "image_url",
+                "image_url": {"url": image_url},
+                "frame_type": "first_frame",
+            }
+        ],
     }
     if duration_seconds is not None:
-        payload["duration_seconds"] = duration_seconds
+        payload["duration"] = duration_seconds
     if resolution is not None:
         payload["resolution"] = resolution
+    if generate_audio is not None:
+        payload["generate_audio"] = generate_audio
     async with httpx.AsyncClient(timeout=120) as client:
         resp = client.build_request(
             "POST", f"{base_url}/videos", headers=_headers(), json=payload
@@ -141,6 +170,18 @@ async def poll_video_status(polling_url: str) -> dict[str, Any]:
         return response.json()
 
 
+async def list_video_models() -> list[dict[str, Any]]:
+    """Return video generation models from the dedicated /videos/models endpoint."""
+    base_url = os.getenv("OPENROUTER_BASE_URL", OPENROUTER_BASE_URL)
+    async with httpx.AsyncClient(timeout=15) as client:
+        resp = client.build_request(
+            "GET", f"{base_url}/videos/models", headers=_headers()
+        )
+        response = await client.send(resp)
+        response.raise_for_status()
+        return response.json().get("data", [])
+
+
 async def generate_image_chat(
     model: str,
     prompt: str,

backend/app/services/video_worker.py

@@ -0,0 +1,159 @@
+"""Background worker: processes queued/processing video generation jobs."""
+import asyncio
+import json
+import logging
+from datetime import datetime, timezone
+
+import duckdb
+
+from . import openrouter
+from .models import mark_timed_out_video_jobs
+
+logger = logging.getLogger(__name__)
+
+# Interval between worker ticks (seconds)
+WORKER_INTERVAL = 15
+# Jobs to process per tick (prevents unbounded bursts)
+BATCH_SIZE = 5
+
+
+async def process_queued_jobs(conn: duckdb.DuckDBPyConnection, lock: asyncio.Lock) -> int:
+    """Submit queued jobs to OpenRouter and transition them to 'processing'."""
+    rows = conn.execute(
+        """SELECT id, generation_type, request_params
+           FROM generated_videos
+           WHERE status = 'queued' AND request_params IS NOT NULL
+           ORDER BY created_at ASC
+           LIMIT ?""",
+        [BATCH_SIZE],
+    ).fetchall()
+
+    processed = 0
+    for row in rows:
+        db_id, generation_type, raw_params = str(row[0]), row[1], row[2]
+        try:
+            params = json.loads(raw_params)
+        except (json.JSONDecodeError, TypeError):
+            logger.error("Bad request_params for video job %s", db_id)
+            continue
+
+        try:
+            if generation_type == "image_to_video":
+                result = await openrouter.generate_video_from_image(
+                    model=params["model"],
+                    image_url=params.get("image_url", ""),
+                    prompt=params.get("prompt", ""),
+                    duration_seconds=params.get("duration_seconds"),
+                    aspect_ratio=params.get("aspect_ratio", "16:9"),
+                    resolution=params.get("resolution"),
+                )
+            else:
+                result = await openrouter.generate_video(
+                    model=params["model"],
+                    prompt=params.get("prompt", ""),
+                    duration_seconds=params.get("duration_seconds"),
+                    aspect_ratio=params.get("aspect_ratio", "16:9"),
+                    resolution=params.get("resolution"),
+                )
+        except Exception as exc:
+            logger.warning("OpenRouter call failed for job %s: %s", db_id, exc)
+            now = datetime.now(timezone.utc).replace(tzinfo=None)
+            async with lock:
+                conn.execute(
+                    "UPDATE generated_videos SET status = 'failed', error = ?, updated_at = ? WHERE id = ?",
+                    [str(exc), now, db_id],
+                )
+            continue
+
+        job_id = result.get("id", "")
+        polling_url = result.get("polling_url")
+        new_status = result.get("status", "processing")
+        # Normalise terminal statuses returned immediately (rare but possible)
+        if new_status not in ("queued", "processing", "completed", "failed", "cancelled"):
+            new_status = "processing"
+
+        urls = result.get("unsigned_urls") or result.get("video_urls")
+        video_url = (urls or [None])[0]
+        now = datetime.now(timezone.utc).replace(tzinfo=None)
+
+        async with lock:
+            conn.execute(
+                """UPDATE generated_videos
+                   SET job_id = ?, polling_url = ?, status = ?, video_url = ?, updated_at = ?
+                   WHERE id = ?""",
+                [job_id, polling_url, new_status, video_url, now, db_id],
+            )
+        processed += 1
+        logger.info("Video job %s → %s (provider id: %s)",
+                    db_id, new_status, job_id)
+
+    return processed
+
+
+async def process_processing_jobs(conn: duckdb.DuckDBPyConnection, lock: asyncio.Lock) -> int:
+    """Poll in-progress jobs and update to 'completed' or 'failed'."""
+    rows = conn.execute(
+        """SELECT id, polling_url
+           FROM generated_videos
+           WHERE status = 'processing' AND polling_url IS NOT NULL
+           ORDER BY updated_at ASC
+           LIMIT ?""",
+        [BATCH_SIZE],
+    ).fetchall()
+
+    updated = 0
+    for row in rows:
+        db_id, polling_url = str(row[0]), row[1]
+        try:
+            result = await openrouter.poll_video_status(polling_url)
+        except Exception as exc:
+            logger.warning("Polling failed for job %s: %s", db_id, exc)
+            continue
+
+        job_status = result.get("status", "processing")
+        if job_status not in ("completed", "failed"):
+            continue  # still in-progress — check again next tick
+
+        urls = result.get("unsigned_urls") or result.get("video_urls")
+        video_url = (urls or [None])[0]
+        error_msg = result.get("error")
+        now = datetime.now(timezone.utc).replace(tzinfo=None)
+
+        async with lock:
+            conn.execute(
+                """UPDATE generated_videos
+                   SET status = ?, video_url = ?, error = ?, updated_at = ?
+                   WHERE id = ?""",
+                [job_status, video_url, error_msg, now, db_id],
+            )
+        updated += 1
+        logger.info("Video job %s → %s", db_id, job_status)
+
+    return updated
+
+
+async def worker_tick(conn: duckdb.DuckDBPyConnection, lock: asyncio.Lock) -> None:
+    """Single worker tick: submit queued, poll processing, expire timed-out."""
+    queued = await process_queued_jobs(conn, lock)
+    polled = await process_processing_jobs(conn, lock)
+    async with lock:
+        timed_out = mark_timed_out_video_jobs(conn, timeout_minutes=120)
+    if queued or polled or timed_out:
+        logger.info(
+            "Worker tick: submitted=%d polled=%d timed_out=%d",
+            queued, polled, timed_out,
+        )
+
+
+async def run_worker(conn: duckdb.DuckDBPyConnection, lock: asyncio.Lock) -> None:
+    """Infinite loop: run a worker tick every WORKER_INTERVAL seconds."""
+    logger.info("Video worker started (interval=%ds)", WORKER_INTERVAL)
+    while True:
+        try:
+            await worker_tick(conn, lock)
+        except asyncio.CancelledError:
+            logger.info("Video worker stopped.")
+            return
+        except Exception as exc:
+            logger.exception("Unexpected error in video worker: %s", exc)
+        await asyncio.sleep(WORKER_INTERVAL)

backend/requirements.txt

@@ -1,5 +1,5 @@
 anyio
-bcrypt
+bcrypt==4.0.1
 blinker
 certifi
 cryptography
@@ -13,8 +13,9 @@ httpx
 Jinja2
 MarkupSafe
 packaging
-passlib
+passlib==1.7.4
 pydantic
 python-dotenv
 python-jose
+python-multipart
 uvicorn

backend/tests/test_admin.py

@@ -53,7 +53,8 @@ async def test_stats_as_admin(client):
     resp = await client.get("/admin/stats", headers={"Authorization": f"Bearer {token}"})
     assert resp.status_code == 200
     data = resp.json()
-    assert data["users"]["total"] == 3  # 2 users + 1 admin
+    # 2 users + 1 admin + 1 seeded admin (ai@allucanget.biz)
+    assert data["users"]["total"] == 4
     assert "by_role" in data["users"]
     assert "refresh_tokens" in data
 

backend/tests/test_ai.py

@@ -53,7 +53,7 @@ async def _user_token(client):
 async def test_list_models(client):
     token = await _user_token(client)
     with patch(
-        "backend.app.routers.ai.openrouter.list_models",
+        "app.routers.ai.openrouter.list_models",
         new_callable=AsyncMock,
         return_value=FAKE_MODELS,
     ):
@@ -74,7 +74,7 @@ async def test_list_models_unauthenticated(client):
 async def test_list_models_upstream_error(client):
     token = await _user_token(client)
     with patch(
-        "backend.app.routers.ai.openrouter.list_models",
+        "app.routers.ai.openrouter.list_models",
         new_callable=AsyncMock,
         side_effect=Exception("Connection refused"),
     ):
@@ -91,7 +91,7 @@ async def test_list_models_upstream_error(client):
 async def test_chat_success(client):
     token = await _user_token(client)
     with patch(
-        "backend.app.routers.ai.openrouter.chat_completion",
+        "app.routers.ai.openrouter.chat_completion",
         new_callable=AsyncMock,
         return_value=FAKE_CHAT_RESPONSE,
     ):
@@ -115,7 +115,7 @@ async def test_chat_success(client):
 async def test_chat_passes_parameters(client):
     token = await _user_token(client)
     mock = AsyncMock(return_value=FAKE_CHAT_RESPONSE)
-    with patch("backend.app.routers.ai.openrouter.chat_completion", new_callable=AsyncMock, return_value=FAKE_CHAT_RESPONSE) as mock:
+    with patch("app.routers.ai.openrouter.chat_completion", new_callable=AsyncMock, return_value=FAKE_CHAT_RESPONSE) as mock:
         await client.post(
             "/ai/chat",
             json={
@@ -145,7 +145,7 @@ async def test_chat_unauthenticated(client):
 async def test_chat_upstream_error(client):
     token = await _user_token(client)
     with patch(
-        "backend.app.routers.ai.openrouter.chat_completion",
+        "app.routers.ai.openrouter.chat_completion",
         new_callable=AsyncMock,
         side_effect=Exception("timeout"),
     ):
@@ -160,7 +160,7 @@ async def test_chat_upstream_error(client):
 async def test_chat_malformed_upstream_response(client):
     token = await _user_token(client)
     with patch(
-        "backend.app.routers.ai.openrouter.chat_completion",
+        "app.routers.ai.openrouter.chat_completion",
         new_callable=AsyncMock,
         return_value={"id": "x", "choices": []},  # empty choices
     ):

backend/tests/test_db.py

@@ -192,3 +192,39 @@ async def test_write_lock_serialises_concurrent_writes():
     # Each writer's start and end must be adjacent (no interleaving)
     assert order.index("A-start") + 1 == order.index("A-end") or \
         order.index("B-start") + 1 == order.index("B-end")
+
+
+# ---------------------------------------------------------------------------
+# Admin seed user
+# ---------------------------------------------------------------------------
+
+def test_seed_admin_user_created_on_init():
+    import os
+    conn = db_module.init_db(":memory:")
+    row = conn.execute(
+        "SELECT email, role FROM users WHERE email = 'ai@allucanget.biz'"
+    ).fetchone()
+    assert row is not None
+    assert row[0] == "ai@allucanget.biz"
+    assert row[1] == "admin"
+
+
+def test_seed_admin_is_idempotent():
+    conn = db_module.init_db(":memory:")
+    # Simulate re-running seed (second init_db call reuses connection, so call _seed_admin directly)
+    db_module._seed_admin(conn)
+    count = conn.execute(
+        "SELECT COUNT(*) FROM users WHERE email = 'ai@allucanget.biz'"
+    ).fetchone()[0]
+    assert count == 1
+
+
+def test_seed_admin_email_env_override(monkeypatch):
+    monkeypatch.setenv("ADMIN_EMAIL", "custom@example.com")
+    monkeypatch.setenv("ADMIN_PASSWORD", "custompass")
+    conn = db_module.init_db(":memory:")
+    row = conn.execute(
+        "SELECT email, role FROM users WHERE email = 'custom@example.com'"
+    ).fetchone()
+    assert row is not None
+    assert row[1] == "admin"

backend/tests/test_generate.py

@@ -18,15 +18,6 @@ FAKE_CHAT = {
     "usage": {"prompt_tokens": 5, "completion_tokens": 10, "total_tokens": 15},
 }
 
-FAKE_IMAGE = {
-    "id": "gen-img-1",
-    "model": "openai/dall-e-3",
-    "data": [
-        {"url": "https://example.com/image.png",
-            "revised_prompt": "A cat on the moon"},
-    ],
-}
-
 FAKE_VIDEO = {
     "id": "gen-vid-1",
     "polling_url": "https://openrouter.ai/api/v1/videos/gen-vid-1",
@@ -69,7 +60,7 @@ async def _user_token(client):
 
 async def test_generate_text(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.chat_completion", new_callable=AsyncMock, return_value=FAKE_CHAT):
+    with patch("app.routers.generate.openrouter.chat_completion", new_callable=AsyncMock, return_value=FAKE_CHAT):
         resp = await client.post(
             "/generate/text",
             json={"model": "openai/gpt-4o", "prompt": "Tell me a story"},
@@ -85,7 +76,7 @@ async def test_generate_text(client):
 async def test_generate_text_with_system_prompt(client):
     token = await _user_token(client)
     mock = AsyncMock(return_value=FAKE_CHAT)
-    with patch("backend.app.routers.generate.openrouter.chat_completion", mock):
+    with patch("app.routers.generate.openrouter.chat_completion", mock):
         await client.post(
             "/generate/text",
             json={"model": "openai/gpt-4o", "prompt": "Hello",
@@ -97,6 +88,44 @@ async def test_generate_text_with_system_prompt(client):
     assert call_messages[1] == {"role": "user", "content": "Hello"}
 
 
+async def test_generate_text_with_messages_array(client):
+    """messages field takes precedence over prompt for multi-turn chat."""
+    token = await _user_token(client)
+    mock = AsyncMock(return_value=FAKE_CHAT)
+    messages = [
+        {"role": "user", "content": "First message"},
+        {"role": "assistant", "content": "Reply"},
+        {"role": "user", "content": "Follow up"},
+    ]
+    with patch("app.routers.generate.openrouter.chat_completion", mock):
+        resp = await client.post(
+            "/generate/text",
+            json={"model": "openai/gpt-4o", "messages": messages},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    assert resp.status_code == 200
+    call_messages = mock.call_args.kwargs["messages"]
+    assert len(call_messages) == 3
+    assert call_messages[2]["content"] == "Follow up"
+
+
+async def test_generate_text_messages_with_system_prompt(client):
+    """system_prompt prepended when messages provided and no system msg present."""
+    token = await _user_token(client)
+    mock = AsyncMock(return_value=FAKE_CHAT)
+    messages = [{"role": "user", "content": "Hi"}]
+    with patch("app.routers.generate.openrouter.chat_completion", mock):
+        await client.post(
+            "/generate/text",
+            json={"model": "openai/gpt-4o", "messages": messages,
+                  "system_prompt": "Be brief."},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    call_messages = mock.call_args.kwargs["messages"]
+    assert call_messages[0] == {"role": "system", "content": "Be brief."}
+    assert call_messages[1] == {"role": "user", "content": "Hi"}
+
+
 async def test_generate_text_unauthenticated(client):
     resp = await client.post("/generate/text", json={"model": "openai/gpt-4o", "prompt": "Hi"})
     assert resp.status_code == 401
@@ -104,7 +133,7 @@ async def test_generate_text_unauthenticated(client):
 
 async def test_generate_text_upstream_error(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.chat_completion", new_callable=AsyncMock, side_effect=Exception("timeout")):
+    with patch("app.routers.generate.openrouter.chat_completion", new_callable=AsyncMock, side_effect=Exception("timeout")):
         resp = await client.post(
             "/generate/text",
             json={"model": "openai/gpt-4o", "prompt": "Hi"},
@@ -117,47 +146,13 @@ async def test_generate_text_upstream_error(client):
 # POST /generate/image
 # ---------------------------------------------------------------------------
 
-async def test_generate_image(client):
-    token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_image", new_callable=AsyncMock, return_value=FAKE_IMAGE):
-        resp = await client.post(
-            "/generate/image",
-            json={"model": "openai/dall-e-3", "prompt": "A cat on the moon"},
-            headers={"Authorization": f"Bearer {token}"},
-        )
-    assert resp.status_code == 200
-    data = resp.json()
-    assert data["id"] == "gen-img-1"
-    assert len(data["images"]) == 1
-    assert data["images"][0]["url"] == "https://example.com/image.png"
-    assert data["images"][0]["revised_prompt"] == "A cat on the moon"
-
-
-async def test_generate_image_unauthenticated(client):
-    resp = await client.post("/generate/image", json={"model": "openai/dall-e-3", "prompt": "Hi"})
-    assert resp.status_code == 401
-
-
-async def test_generate_image_upstream_error(client):
-    token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_image", new_callable=AsyncMock, side_effect=Exception("rate limit")):
-        resp = await client.post(
-            "/generate/image",
-            json={"model": "openai/dall-e-3", "prompt": "Hi"},
-            headers={"Authorization": f"Bearer {token}"},
-        )
-    assert resp.status_code == 502
-
-
-# --- Chat-based image generation (FLUX, GPT-5 Image Mini) ---
-
 FAKE_IMAGE_CHAT_FLUX = {
     "id": "gen-img-chat-1",
     "model": "black-forest-labs/flux.2-klein-4b",
     "choices": [{
         "message": {
             "role": "assistant",
-            "content": "Here is your generated image.",
+            "content": None,
             "images": [{
                 "type": "image_url",
                 "image_url": {"url": "data:image/png;base64,abc123"},
@@ -181,45 +176,65 @@ FAKE_IMAGE_CHAT_GPT5 = {
     }],
 }
 
+FAKE_IMAGE_CHAT_GEMINI = {
+    "id": "gen-img-chat-3",
+    "model": "google/gemini-2.5-flash-image",
+    "choices": [{
+        "message": {
+            "role": "assistant",
+            "content": "Here is your image.",
+            "images": [{
+                "type": "image_url",
+                "image_url": {"url": "data:image/png;base64,gemini123"},
+            }],
+        }
+    }],
+}
 
-async def test_generate_image_chat_flux(client):
+
+async def test_generate_image(client):
+    """All models now use generate_image_chat (chat completions endpoint)."""
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_image_chat", new_callable=AsyncMock, return_value=FAKE_IMAGE_CHAT_FLUX):
+    with patch("app.routers.generate.openrouter.generate_image_chat", new_callable=AsyncMock, return_value=FAKE_IMAGE_CHAT_GEMINI):
         resp = await client.post(
             "/generate/image",
-            json={"model": "black-forest-labs/flux.2-klein-4b",
-                  "prompt": "A sunset"},
+            json={"model": "google/gemini-2.5-flash-image",
+                  "prompt": "A cat on the moon"},
             headers={"Authorization": f"Bearer {token}"},
         )
     assert resp.status_code == 200
     data = resp.json()
-    assert data["id"] == "gen-img-chat-1"
+    assert data["id"] == "gen-img-chat-3"
     assert len(data["images"]) == 1
-    assert data["images"][0]["url"] == "data:image/png;base64,abc123"
+    assert data["images"][0]["url"] == "data:image/png;base64,gemini123"
+    assert data["images"][0]["image_id"] is not None  # stored in DB
 
 
-async def test_generate_image_chat_gpt5_image_mini(client):
+async def test_generate_image_unauthenticated(client):
+    resp = await client.post("/generate/image", json={"model": "google/gemini-2.5-flash-image", "prompt": "Hi"})
+    assert resp.status_code == 401
+
+
+async def test_generate_image_upstream_error(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_image_chat", new_callable=AsyncMock, return_value=FAKE_IMAGE_CHAT_GPT5):
+    with patch("app.routers.generate.openrouter.generate_image_chat", new_callable=AsyncMock, side_effect=Exception("rate limit")):
         resp = await client.post(
             "/generate/image",
-            json={"model": "openai/gpt-5-image-mini", "prompt": "A cat"},
+            json={"model": "google/gemini-2.5-flash-image", "prompt": "Hi"},
             headers={"Authorization": f"Bearer {token}"},
         )
-    assert resp.status_code == 200
-    data = resp.json()
-    assert data["model"] == "openai/gpt-5-image-mini"
-    assert len(data["images"]) == 1
+    assert resp.status_code == 502
 
 
-async def test_generate_image_chat_with_image_config(client):
+async def test_generate_image_with_image_config(client):
+    """Passes aspect_ratio + image_size through to generate_image_chat."""
     token = await _user_token(client)
-    mock = AsyncMock(return_value=FAKE_IMAGE_CHAT_FLUX)
-    with patch("backend.app.routers.generate.openrouter.generate_image_chat", mock):
+    mock = AsyncMock(return_value=FAKE_IMAGE_CHAT_GEMINI)
+    with patch("app.routers.generate.openrouter.generate_image_chat", mock):
         await client.post(
             "/generate/image",
             json={
-                "model": "black-forest-labs/flux.2-klein-4b",
+                "model": "google/gemini-2.5-flash-image",
                 "prompt": "A landscape",
                 "aspect_ratio": "16:9",
                 "image_size": "2K",
@@ -229,23 +244,112 @@ async def test_generate_image_chat_with_image_config(client):
     call_kwargs = mock.call_args.kwargs
     assert call_kwargs["image_config"]["aspect_ratio"] == "16:9"
     assert call_kwargs["image_config"]["image_size"] == "2K"
-    assert call_kwargs["modalities"] == ["image"]
 
 
-async def test_generate_image_chat_unauthenticated(client):
-    resp = await client.post("/generate/image", json={"model": "flux.2-klein-4b", "prompt": "Hi"})
-    assert resp.status_code == 401
-
-
-async def test_generate_image_chat_upstream_error(client):
+async def test_generate_image_default_modalities_image_text(client):
+    """Model not in cache → default modalities = ['image', 'text']."""
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_image_chat", new_callable=AsyncMock, side_effect=Exception("timeout")):
+    mock = AsyncMock(return_value=FAKE_IMAGE_CHAT_GEMINI)
+    with patch("app.routers.generate.openrouter.generate_image_chat", mock):
+        await client.post(
+            "/generate/image",
+            json={"model": "google/gemini-2.5-flash-image", "prompt": "Hi"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    assert mock.call_args.kwargs["modalities"] == ["image", "text"]
+
+
+async def test_generate_image_image_only_modalities_from_cache(client):
+    """Model cached with image-only output_modalities → modalities = ['image']."""
+    from app import db as db_module
+    from app.services.models import get_model_output_modalities
+    import json as _json
+    token = await _user_token(client)
+
+    # Seed cache with image-only model
+    conn = db_module.get_conn()
+    from datetime import datetime, timezone
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+    conn.execute(
+        "DELETE FROM models_cache WHERE model_id = 'black-forest-labs/flux.2-pro'"
+    )
+    conn.execute(
+        """INSERT INTO models_cache (model_id, name, modality, context_length, pricing, fetched_at, output_modalities)
+           VALUES (?, ?, ?, ?, ?, ?, ?)""",
+        ["black-forest-labs/flux.2-pro", "FLUX.2 Pro", "image", None, None, now,
+         _json.dumps(["image"])],
+    )
+
+    mock = AsyncMock(return_value=FAKE_IMAGE_CHAT_FLUX)
+    with patch("app.routers.generate.openrouter.generate_image_chat", mock):
         resp = await client.post(
             "/generate/image",
-            json={"model": "black-forest-labs/flux.2-klein-4b", "prompt": "Hi"},
+            json={"model": "black-forest-labs/flux.2-pro", "prompt": "Sky"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    assert resp.status_code == 200
+    assert mock.call_args.kwargs["modalities"] == ["image"]
+
+
+async def test_generate_image_no_images_in_response(client):
+    """502 when model returns no images."""
+    token = await _user_token(client)
+    empty_response = {
+        "id": "gen-empty",
+        "model": "google/gemini-2.5-flash-image",
+        "choices": [{"message": {"role": "assistant", "content": "ok", "images": []}}],
+    }
+    with patch("app.routers.generate.openrouter.generate_image_chat",
+               new_callable=AsyncMock, return_value=empty_response):
+        resp = await client.post(
+            "/generate/image",
+            json={"model": "google/gemini-2.5-flash-image", "prompt": "Hi"},
             headers={"Authorization": f"Bearer {token}"},
         )
     assert resp.status_code == 502
+    assert "No images returned" in resp.json()["detail"]
+
+
+async def test_generate_image_flux(client):
+    """Flux model works correctly via chat completions."""
+    token = await _user_token(client)
+    with patch("app.routers.generate.openrouter.generate_image_chat",
+               new_callable=AsyncMock, return_value=FAKE_IMAGE_CHAT_FLUX):
+        resp = await client.post(
+            "/generate/image",
+            json={"model": "black-forest-labs/flux.2-klein-4b",
+                  "prompt": "A sunset"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    assert resp.status_code == 200
+    data = resp.json()
+    assert data["images"][0]["url"] == "data:image/png;base64,abc123"
+
+
+async def test_generate_image_stored_in_db(client):
+    """Generated image row persists in generated_images table."""
+    from app import db as db_module
+    token = await _user_token(client)
+    with patch("app.routers.generate.openrouter.generate_image_chat",
+               new_callable=AsyncMock, return_value=FAKE_IMAGE_CHAT_GEMINI):
+        resp = await client.post(
+            "/generate/image",
+            json={"model": "google/gemini-2.5-flash-image",
+                  "prompt": "A mountain"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    assert resp.status_code == 200
+    image_id = resp.json()["images"][0]["image_id"]
+    assert image_id is not None
+
+    row = db_module.get_conn().execute(
+        "SELECT model_id, prompt, image_data FROM generated_images WHERE id = ?",
+        [image_id],
+    ).fetchone()
+    assert row is not None
+    assert row[0] == "google/gemini-2.5-flash-image"
+    assert row[1] == "A mountain"
+    assert row[2] == "data:image/png;base64,gemini123"
 
 
 # ---------------------------------------------------------------------------
@@ -254,7 +358,7 @@ async def test_generate_image_chat_upstream_error(client):
 
 async def test_generate_video(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_video", new_callable=AsyncMock, return_value=FAKE_VIDEO):
+    with patch("app.routers.generate.openrouter.generate_video", new_callable=AsyncMock, return_value=FAKE_VIDEO):
         resp = await client.post(
             "/generate/video",
             json={"model": "stability/stable-video",
@@ -276,7 +380,7 @@ async def test_generate_video_unauthenticated(client):
 
 async def test_generate_video_upstream_error(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_video", new_callable=AsyncMock, side_effect=Exception("503")):
+    with patch("app.routers.generate.openrouter.generate_video", new_callable=AsyncMock, side_effect=Exception("503")):
         resp = await client.post(
             "/generate/video",
             json={"model": "stability/stable-video", "prompt": "Hi"},
@@ -291,7 +395,7 @@ async def test_generate_video_upstream_error(client):
 
 async def test_generate_video_from_image(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_video_from_image", new_callable=AsyncMock, return_value=FAKE_VIDEO_DONE):
+    with patch("app.routers.generate.openrouter.generate_video_from_image", new_callable=AsyncMock, return_value=FAKE_VIDEO_DONE):
         resp = await client.post(
             "/generate/video/from-image",
             json={
@@ -315,7 +419,7 @@ async def test_poll_video_status(client):
         "status": "completed",
         "unsigned_urls": ["https://example.com/video.mp4"],
     }
-    with patch("backend.app.routers.generate.openrouter.poll_video_status", new_callable=AsyncMock, return_value=mock_result):
+    with patch("app.routers.generate.openrouter.poll_video_status", new_callable=AsyncMock, return_value=mock_result):
         resp = await client.get(
             "/generate/video/status",
             params={"polling_url": "https://openrouter.ai/api/v1/videos/gen-vid-1"},
@@ -337,7 +441,7 @@ async def test_poll_video_status_unauthenticated(client):
 
 async def test_poll_video_status_upstream_error(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.poll_video_status", new_callable=AsyncMock, side_effect=Exception("timeout")):
+    with patch("app.routers.generate.openrouter.poll_video_status", new_callable=AsyncMock, side_effect=Exception("timeout")):
         resp = await client.get(
             "/generate/video/status",
             params={"polling_url": "https://openrouter.ai/api/v1/videos/gen-vid-1"},
@@ -354,9 +458,130 @@ async def test_generate_video_from_image_unauthenticated(client):
     assert resp.status_code == 401
 
 
+# ---------------------------------------------------------------------------
+# Video job DB storage
+# ---------------------------------------------------------------------------
+
+async def test_generate_video_stored_in_db(client):
+    """Submitting a video job inserts a row into generated_videos."""
+    token = await _user_token(client)
+    with patch("app.routers.generate.openrouter.generate_video", new_callable=AsyncMock, return_value=FAKE_VIDEO):
+        resp = await client.post(
+            "/generate/video",
+            json={"model": "stability/stable-video", "prompt": "Ocean waves"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    assert resp.status_code == 200
+
+    row = db_module.get_conn().execute(
+        "SELECT job_id, model_id, prompt, status FROM generated_videos WHERE job_id = ?",
+        ["gen-vid-1"],
+    ).fetchone()
+    assert row is not None
+    assert row[0] == "gen-vid-1"
+    assert row[1] == "stability/stable-video"
+    assert row[2] == "Ocean waves"
+    assert row[3] == "queued"
+
+
+async def test_generate_video_from_image_stored_in_db(client):
+    """Submitting a from-image job inserts a row into generated_videos."""
+    token = await _user_token(client)
+    with patch("app.routers.generate.openrouter.generate_video_from_image", new_callable=AsyncMock, return_value=FAKE_VIDEO_DONE):
+        resp = await client.post(
+            "/generate/video/from-image",
+            json={
+                "model": "runway/gen-3",
+                "image_url": "https://example.com/cat.jpg",
+                "prompt": "Cat runs",
+            },
+            headers={"Authorization": f"Bearer {token}"},
+        )
+    assert resp.status_code == 200
+
+    row = db_module.get_conn().execute(
+        "SELECT job_id, model_id, prompt, status FROM generated_videos WHERE job_id = ?",
+        ["gen-vid-2"],
+    ).fetchone()
+    assert row is not None
+    assert row[1] == "runway/gen-3"
+    assert row[2] == "Cat runs"
+
+
+async def test_poll_video_updates_db_on_completion(client):
+    """Polling a completed job updates the row status and video_url."""
+    token = await _user_token(client)
+    # First submit a job
+    with patch("app.routers.generate.openrouter.generate_video", new_callable=AsyncMock, return_value=FAKE_VIDEO):
+        await client.post(
+            "/generate/video",
+            json={"model": "stability/stable-video", "prompt": "Test"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+
+    # Now poll and get completed status
+    mock_result = {
+        "id": "gen-vid-1",
+        "status": "completed",
+        "unsigned_urls": ["https://example.com/video.mp4"],
+    }
+    with patch("app.routers.generate.openrouter.poll_video_status", new_callable=AsyncMock, return_value=mock_result):
+        await client.get(
+            "/generate/video/status",
+            params={"polling_url": "https://openrouter.ai/api/v1/videos/gen-vid-1"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+
+    row = db_module.get_conn().execute(
+        "SELECT status, video_url FROM generated_videos WHERE job_id = ?",
+        ["gen-vid-1"],
+    ).fetchone()
+    assert row is not None
+    assert row[0] == "completed"
+    assert row[1] == "https://example.com/video.mp4"
+
+
+async def test_list_generated_videos_empty(client):
+    """GET /generate/videos returns empty list initially."""
+    token = await _user_token(client)
+    resp = await client.get(
+        "/generate/videos",
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    assert resp.status_code == 200
+    assert resp.json() == []
+
+
+async def test_list_generated_videos_returns_own_jobs(client):
+    """GET /generate/videos returns the current user's jobs only."""
+    token = await _user_token(client)
+    with patch("app.routers.generate.openrouter.generate_video", new_callable=AsyncMock, return_value=FAKE_VIDEO):
+        await client.post(
+            "/generate/video",
+            json={"model": "stability/stable-video", "prompt": "Waves"},
+            headers={"Authorization": f"Bearer {token}"},
+        )
+
+    resp = await client.get(
+        "/generate/videos",
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    assert resp.status_code == 200
+    data = resp.json()
+    assert len(data) == 1
+    assert data[0]["job_id"] == "gen-vid-1"
+    assert data[0]["prompt"] == "Waves"
+    assert data[0]["status"] == "queued"
+
+
+async def test_list_generated_videos_unauthenticated(client):
+    resp = await client.get("/generate/videos")
+    assert resp.status_code == 401
+
+
 async def test_generate_video_from_image_upstream_error(client):
     token = await _user_token(client)
-    with patch("backend.app.routers.generate.openrouter.generate_video_from_image", new_callable=AsyncMock, side_effect=Exception("error")):
+    with patch("app.routers.generate.openrouter.generate_video_from_image", new_callable=AsyncMock, side_effect=Exception("error")):
         resp = await client.post(
             "/generate/video/from-image",
             json={"model": "runway/gen-3",

backend/tests/test_images.py

@@ -0,0 +1,184 @@
+"""Tests for image upload and retrieval endpoints."""
+import io
+import os
+import pytest
+import pytest_asyncio
+from httpx import AsyncClient, ASGITransport
+
+from app.main import app
+from app import db as db_module
+
+os.environ.setdefault("JWT_SECRET", "test-secret-key-for-testing-only")
+# Use a temp dir so file I/O works without polluting project data/
+os.environ.setdefault("UPLOAD_DIR", "/tmp/test_uploads")
+
+
+@pytest.fixture(autouse=True)
+def fresh_db():
+    db_module._conn = None
+    db_module.init_db(":memory:")
+    yield
+    db_module.close_db()
+    db_module._conn = None
+
+
+@pytest_asyncio.fixture
+async def client(fresh_db):
+    transport = ASGITransport(app=app)
+    async with AsyncClient(transport=transport, base_url="http://test") as ac:
+        yield ac
+
+
+async def _user_token(client) -> str:
+    await client.post("/auth/register", json={"email": "user@example.com", "password": "secret123"})
+    resp = await client.post("/auth/login", json={"email": "user@example.com", "password": "secret123"})
+    return resp.json()["access_token"]
+
+
+async def _other_token(client) -> str:
+    await client.post("/auth/register", json={"email": "other@example.com", "password": "secret123"})
+    resp = await client.post("/auth/login", json={"email": "other@example.com", "password": "secret123"})
+    return resp.json()["access_token"]
+
+
+def _png_bytes() -> bytes:
+    """Minimal valid 1x1 PNG."""
+    return (
+        b"\x89PNG\r\n\x1a\n\x00\x00\x00\rIHDR\x00\x00\x00\x01\x00\x00\x00\x01"
+        b"\x08\x02\x00\x00\x00\x90wS\xde\x00\x00\x00\x0cIDATx\x9cc\xf8\x0f\x00"
+        b"\x00\x01\x01\x00\x05\x18\xd4n\x00\x00\x00\x00IEND\xaeB`\x82"
+    )
+
+
+# ---------------------------------------------------------------------------
+# POST /images/upload
+# ---------------------------------------------------------------------------
+
+async def test_upload_image_success(client):
+    token = await _user_token(client)
+    resp = await client.post(
+        "/images/upload",
+        files={"file": ("test.png", io.BytesIO(_png_bytes()), "image/png")},
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    assert resp.status_code == 201
+    data = resp.json()
+    assert data["filename"] == "test.png"
+    assert data["content_type"] == "image/png"
+    assert "id" in data
+    assert data["size_bytes"] > 0
+
+
+async def test_upload_image_unauthenticated(client):
+    resp = await client.post(
+        "/images/upload",
+        files={"file": ("test.png", io.BytesIO(_png_bytes()), "image/png")},
+    )
+    assert resp.status_code == 401
+
+
+async def test_upload_image_unsupported_type(client):
+    token = await _user_token(client)
+    resp = await client.post(
+        "/images/upload",
+        files={"file": ("doc.pdf", io.BytesIO(
+            b"%PDF-fake"), "application/pdf")},
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    assert resp.status_code == 415
+
+
+async def test_upload_image_too_large(client, monkeypatch):
+    import app.routers.images as images_mod
+    monkeypatch.setattr(images_mod, "MAX_SIZE_BYTES", 5)
+    token = await _user_token(client)
+    resp = await client.post(
+        "/images/upload",
+        files={"file": ("big.png", io.BytesIO(b"x" * 10), "image/png")},
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    assert resp.status_code == 413
+
+
+# ---------------------------------------------------------------------------
+# GET /images/
+# ---------------------------------------------------------------------------
+
+async def test_list_images_empty(client):
+    token = await _user_token(client)
+    resp = await client.get("/images/", headers={"Authorization": f"Bearer {token}"})
+    assert resp.status_code == 200
+    assert resp.json() == []
+
+
+async def test_list_images_returns_own_only(client):
+    token = await _user_token(client)
+    other = await _other_token(client)
+
+    # Upload one image as user, one as other
+    for tok, name in [(token, "mine.png"), (other, "theirs.png")]:
+        await client.post(
+            "/images/upload",
+            files={"file": (name, io.BytesIO(_png_bytes()), "image/png")},
+            headers={"Authorization": f"Bearer {tok}"},
+        )
+
+    resp = await client.get("/images/", headers={"Authorization": f"Bearer {token}"})
+    assert resp.status_code == 200
+    items = resp.json()
+    assert len(items) == 1
+    assert items[0]["filename"] == "mine.png"
+
+
+async def test_list_images_unauthenticated(client):
+    resp = await client.get("/images/")
+    assert resp.status_code == 401
+
+
+# ---------------------------------------------------------------------------
+# GET /images/{id}/file
+# ---------------------------------------------------------------------------
+
+async def test_serve_image_success(client):
+    token = await _user_token(client)
+    up = await client.post(
+        "/images/upload",
+        files={"file": ("pixel.png", io.BytesIO(_png_bytes()), "image/png")},
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    image_id = up.json()["id"]
+
+    resp = await client.get(
+        f"/images/{image_id}/file",
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    assert resp.status_code == 200
+    assert resp.headers["content-type"].startswith("image/png")
+    assert resp.content == _png_bytes()
+
+
+async def test_serve_image_wrong_user(client):
+    token = await _user_token(client)
+    other = await _other_token(client)
+
+    up = await client.post(
+        "/images/upload",
+        files={"file": ("secret.png", io.BytesIO(_png_bytes()), "image/png")},
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    image_id = up.json()["id"]
+
+    resp = await client.get(
+        f"/images/{image_id}/file",
+        headers={"Authorization": f"Bearer {other}"},
+    )
+    assert resp.status_code == 403
+
+
+async def test_serve_image_not_found(client):
+    token = await _user_token(client)
+    resp = await client.get(
+        "/images/00000000-0000-0000-0000-000000000000/file",
+        headers={"Authorization": f"Bearer {token}"},
+    )
+    assert resp.status_code == 404

backend/tests/test_models.py

@@ -0,0 +1,366 @@
+"""Tests for model cache service and router."""
+import json
+import os
+from datetime import datetime, timedelta, timezone
+from unittest.mock import AsyncMock, patch
+
+import pytest
+import pytest_asyncio
+from httpx import ASGITransport, AsyncClient
+
+from app import db as db_module
+from app.main import app
+from app.services.models import (
+    _extract_output_modality,
+    _normalize_modality,
+    _parse_modality,
+    get_cached_models,
+    get_model_output_modalities,
+    is_cache_stale,
+    refresh_models_cache,
+)
+
+os.environ.setdefault("JWT_SECRET", "test-secret-key-for-testing-only")
+os.environ.setdefault("OPENROUTER_API_KEY", "test-key")
+
+FAKE_MODELS_RAW = [
+    {
+        "id": "openai/gpt-4o",
+        "name": "GPT-4o",
+        "context_length": 128000,
+        "pricing": {"prompt": "0.000005"},
+        "architecture": {"modality": "text->text", "output_modalities": ["text"]},
+    },
+    {
+        "id": "anthropic/claude-3-haiku",
+        "name": "Claude 3 Haiku",
+        "context_length": 200000,
+        "pricing": {},
+        "architecture": {"modality": "text+image->text", "output_modalities": ["text"]},
+    },
+    {
+        "id": "openai/dall-e-3",
+        "name": "DALL-E 3",
+        "context_length": None,
+        "pricing": {"image": "0.04"},
+        "architecture": {"modality": "text->image", "output_modalities": ["image"]},
+    },
+    {
+        "id": "openai/sora-2",
+        "name": "Sora 2",
+        "context_length": None,
+        "pricing": {"video": "0.10"},
+        "architecture": {"modality": "text->video", "output_modalities": ["video"]},
+    },
+    {
+        "id": "google/gemini-2.5-flash-image",
+        "name": "Gemini 2.5 Flash Image",
+        "context_length": None,
+        "pricing": {},
+        "architecture": {"output_modalities": ["image", "text"]},
+    },
+]
+
+
+@pytest.fixture(autouse=True)
+def fresh_db():
+    db_module._conn = None
+    db_module.init_db(":memory:")
+    yield
+    db_module.close_db()
+    db_module._conn = None
+
+
+@pytest_asyncio.fixture
+async def client(fresh_db):
+    transport = ASGITransport(app=app)
+    async with AsyncClient(transport=transport, base_url="http://test") as ac:
+        yield ac
+
+
+async def _register_login(client, email, password, is_admin=False):
+    """Register + login; optionally promote to admin directly in DB."""
+    await client.post("/auth/register", json={"email": email, "password": password})
+    if is_admin:
+        db_module.get_conn().execute(
+            "UPDATE users SET role = 'admin' WHERE email = ?", [email]
+        )
+    resp = await client.post("/auth/login", json={"email": email, "password": password})
+    return resp.json()["access_token"]
+
+
+# ---------------------------------------------------------------------------
+# Unit tests: _parse_modality
+# ---------------------------------------------------------------------------
+
+def test_parse_modality_text():
+    assert _parse_modality("text->text") == "text"
+
+
+def test_parse_modality_multimodal_input_text_output():
+    assert _parse_modality("text+image->text") == "text"
+
+
+def test_parse_modality_image():
+    assert _parse_modality("text->image") == "image"
+
+
+def test_parse_modality_video():
+    assert _parse_modality("text->video") == "video"
+
+
+def test_parse_modality_audio():
+    assert _parse_modality("text->audio") == "audio"
+
+
+def test_parse_modality_no_arrow_fallback():
+    assert _parse_modality("text") == "text"
+
+
+def test_normalize_embedding_alias():
+    assert _normalize_modality("embedding") == "embeddings"
+
+
+def test_extract_output_modality_prefers_output_modalities():
+    model = {
+        "architecture": {
+            "modality": "text->text",
+            "output_modalities": ["image"],
+        }
+    }
+    assert _extract_output_modality(model) == "image"
+
+
+def test_extract_output_modality_legacy_fallback():
+    model = {"architecture": {"modality": "text->audio"}}
+    assert _extract_output_modality(model) == "audio"
+
+
+# ---------------------------------------------------------------------------
+# Unit tests: is_cache_stale
+# ---------------------------------------------------------------------------
+
+def test_cache_stale_when_empty():
+    conn = db_module.get_conn()
+    assert is_cache_stale(conn) is True
+
+
+def test_cache_not_stale_after_fresh_insert():
+    conn = db_module.get_conn()
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+    conn.execute(
+        "INSERT INTO models_cache (model_id, name, modality, fetched_at) VALUES (?, ?, ?, ?)",
+        ["openai/gpt-4o", "GPT-4o", "text", now],
+    )
+    assert is_cache_stale(conn) is False
+
+
+def test_cache_stale_after_ttl_exceeded():
+    conn = db_module.get_conn()
+    # Store naive UTC to match DuckDB TIMESTAMP behaviour
+    old_time = datetime.now(timezone.utc).replace(
+        tzinfo=None) - timedelta(hours=25)
+    conn.execute(
+        "INSERT INTO models_cache (model_id, name, modality, fetched_at) VALUES (?, ?, ?, ?)",
+        ["openai/gpt-4o", "GPT-4o", "text", old_time],
+    )
+    assert is_cache_stale(conn) is True
+
+
+# ---------------------------------------------------------------------------
+# Unit tests: refresh_models_cache + get_cached_models
+# ---------------------------------------------------------------------------
+
+async def test_refresh_stores_models():
+    conn = db_module.get_conn()
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        return_value=FAKE_MODELS_RAW,
+    ):
+        count = await refresh_models_cache(conn)
+    assert count == 5
+    all_models = get_cached_models(conn)
+    assert len(all_models) == 5
+
+
+async def test_refresh_replaces_old_cache():
+    conn = db_module.get_conn()
+    old_time = datetime.now(timezone.utc).replace(
+        tzinfo=None) - timedelta(hours=30)
+    conn.execute(
+        "INSERT INTO models_cache (model_id, name, modality, fetched_at) VALUES (?, ?, ?, ?)",
+        ["old/model", "Old Model", "text", old_time],
+    )
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        return_value=FAKE_MODELS_RAW,
+    ):
+        await refresh_models_cache(conn)
+    ids = [m["id"] for m in get_cached_models(conn)]
+    assert "old/model" not in ids
+    assert "openai/gpt-4o" in ids
+    assert len(ids) == 5
+
+
+def test_get_cached_models_filter_by_modality():
+    conn = db_module.get_conn()
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+    for m in FAKE_MODELS_RAW:
+        modality = _extract_output_modality(m)
+        conn.execute(
+            "INSERT INTO models_cache (model_id, name, modality, fetched_at) VALUES (?, ?, ?, ?)",
+            [m["id"], m["name"], modality, now],
+        )
+    text_models = get_cached_models(conn, modality="text")
+    # gpt-4o, claude-3-haiku (gemini has output_modalities=["image","text"] → classified as "image")
+    assert len(text_models) == 2
+    assert all(m["modality"] == "text" for m in text_models)
+
+    image_models = get_cached_models(conn, modality="image")
+    # dall-e-3 + gemini (output_modalities starts with image)
+    assert len(image_models) == 2
+    image_ids = [m["id"] for m in image_models]
+    assert "openai/dall-e-3" in image_ids
+    assert "google/gemini-2.5-flash-image" in image_ids
+
+    video_models = get_cached_models(conn, modality="video")
+    assert len(video_models) == 1
+    assert video_models[0]["id"] == "openai/sora-2"
+
+
+# ---------------------------------------------------------------------------
+# Integration tests: GET /models/
+# ---------------------------------------------------------------------------
+
+async def test_list_models_endpoint_auto_refreshes(client):
+    token = await _register_login(client, "user@example.com", "secret123")
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        return_value=FAKE_MODELS_RAW,
+    ) as mock_fetch:
+        resp = await client.get(
+            "/models/", headers={"Authorization": f"Bearer {token}"}
+        )
+    assert resp.status_code == 200
+    assert len(resp.json()) == 5
+    assert mock_fetch.await_count >= 1
+
+
+async def test_list_models_endpoint_uses_cache(client):
+    token = await _register_login(client, "user@example.com", "secret123")
+    conn = db_module.get_conn()
+    now = datetime.now(timezone.utc).replace(tzinfo=None)
+    conn.execute(
+        "INSERT INTO models_cache (model_id, name, modality, fetched_at) VALUES (?, ?, ?, ?)",
+        ["cached/model", "Cached Model", "text", now],
+    )
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+    ) as mock_fetch:
+        resp = await client.get(
+            "/models/?modality=text", headers={"Authorization": f"Bearer {token}"}
+        )
+    assert resp.status_code == 200
+    assert resp.json()[0]["id"] == "cached/model"
+    mock_fetch.assert_not_awaited()
+
+
+async def test_list_models_endpoint_requires_auth(client):
+    resp = await client.get("/models/")
+    assert resp.status_code == 401
+
+
+async def test_list_models_filter_by_modality(client):
+    token = await _register_login(client, "user@example.com", "secret123")
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        return_value=FAKE_MODELS_RAW,
+    ):
+        resp = await client.get(
+            "/models/?modality=image", headers={"Authorization": f"Bearer {token}"}
+        )
+    assert resp.status_code == 200
+    data = resp.json()
+    assert len(data) == 2  # dall-e-3 + gemini-2.5-flash-image
+    image_ids = [m["id"] for m in data]
+    assert "openai/dall-e-3" in image_ids
+    assert "google/gemini-2.5-flash-image" in image_ids
+
+
+# ---------------------------------------------------------------------------
+# Integration tests: POST /models/refresh
+# ---------------------------------------------------------------------------
+
+async def test_refresh_endpoint_requires_admin(client):
+    token = await _register_login(client, "user@example.com", "secret123")
+    resp = await client.post(
+        "/models/refresh", headers={"Authorization": f"Bearer {token}"}
+    )
+    assert resp.status_code == 403
+
+
+async def test_refresh_endpoint_admin_succeeds(client):
+    token = await _register_login(client, "admin@example.com", "secret123", is_admin=True)
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        return_value=FAKE_MODELS_RAW,
+    ):
+        resp = await client.post(
+            "/models/refresh", headers={"Authorization": f"Bearer {token}"}
+        )
+    assert resp.status_code == 200
+    assert resp.json()["refreshed"] == 5
+
+
+async def test_refresh_endpoint_502_on_openrouter_error(client):
+    token = await _register_login(client, "admin@example.com", "secret123", is_admin=True)
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        side_effect=RuntimeError("network error"),
+    ):
+        resp = await client.post(
+            "/models/refresh", headers={"Authorization": f"Bearer {token}"}
+        )
+    assert resp.status_code == 502
+
+
+# ---------------------------------------------------------------------------
+# Unit tests: get_model_output_modalities
+# ---------------------------------------------------------------------------
+
+async def test_get_model_output_modalities_image_only():
+    conn = db_module.get_conn()
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        return_value=FAKE_MODELS_RAW,
+    ):
+        await refresh_models_cache(conn)
+    modalities = get_model_output_modalities(conn, "openai/dall-e-3")
+    assert modalities == ["image"]
+
+
+async def test_get_model_output_modalities_image_text():
+    conn = db_module.get_conn()
+    with patch(
+        "app.services.models.openrouter.list_models",
+        new_callable=AsyncMock,
+        return_value=FAKE_MODELS_RAW,
+    ):
+        await refresh_models_cache(conn)
+    modalities = get_model_output_modalities(
+        conn, "google/gemini-2.5-flash-image")
+    assert set(modalities) == {"image", "text"}
+
+
+def test_get_model_output_modalities_unknown_model():
+    conn = db_module.get_conn()
+    result = get_model_output_modalities(conn, "unknown/model")
+    assert result == []

backend/tests/test_users.py

@@ -115,7 +115,9 @@ async def test_list_users_as_admin(client):
     resp = await client.get("/users", headers={"Authorization": f"Bearer {admin_token}"})
     assert resp.status_code == 200
     assert isinstance(resp.json(), list)
-    assert len(resp.json()) == 1
+    assert len(resp.json()) >= 1
+    emails = [u["email"] for u in resp.json()]
+    assert "user@example.com" in emails
 
 
 async def test_list_users_as_regular_user(client):

docker-compose.coolify.yml

@@ -1,45 +0,0 @@
-services:
-  backend:
-    build:
-      context: ./backend
-      dockerfile: Dockerfile
-    environment:
-      - OPENROUTER_API_KEY=${OPENROUTER_API_KEY}
-      - JWT_SECRET=${JWT_SECRET}
-      - APP_URL=${APP_URL:-https://ai.allucanget.biz}
-      - APP_NAME=${APP_NAME:-AI Allucanget}
-      - CORS_ORIGINS=${CORS_ORIGINS:-https://ai.allucanget.biz}
-    volumes:
-      - app-data:/app/data
-    networks:
-      - app-network
-    healthcheck:
-      test: [ "CMD", "curl", "-f", "http://localhost:12015/health" ]
-      interval: 30s
-      timeout: 10s
-      retries: 3
-      start_period: 5s
-    expose:
-      - "12015"
-
-  frontend:
-    build:
-      context: ./frontend
-      dockerfile: Dockerfile
-    environment:
-      - FLASK_SECRET_KEY=${FLASK_SECRET_KEY}
-      - BACKEND_URL=${BACKEND_URL:-http://backend:12015}
-    depends_on:
-      backend:
-        condition: service_healthy
-    networks:
-      - app-network
-    expose:
-      - "12016"
-
-networks:
-  app-network:
-    driver: bridge
-
-volumes:
-  app-data:

docker-compose.yml

@@ -1,58 +0,0 @@
-services:
-  backend:
-    build:
-      context: ./backend
-      dockerfile: Dockerfile
-    container_name: ai-backend
-    ports:
-      - "12015:12015"
-    environment:
-      - OPENROUTER_API_KEY=${OPENROUTER_API_KEY}
-      - JWT_SECRET=${JWT_SECRET}
-      - APP_URL=${APP_URL:-http://localhost}
-      - APP_NAME=${APP_NAME:-AI Allucanget}
-      - CORS_ORIGINS=${CORS_ORIGINS:-http://localhost:12016}
-    volumes:
-      - ./data:/app/data
-    networks:
-      - app-network
-    healthcheck:
-      test: [ "CMD", "curl", "-f", "http://localhost:12015/health" ]
-      interval: 30s
-      timeout: 10s
-      retries: 3
-      start_period: 5s
-
-  frontend:
-    build:
-      context: ./frontend
-      dockerfile: Dockerfile
-    container_name: ai-frontend
-    ports:
-      - "12016:12016"
-    environment:
-      - FLASK_SECRET_KEY=${FLASK_SECRET_KEY}
-      - BACKEND_URL=${BACKEND_URL:-http://backend:12015}
-    depends_on:
-      backend:
-        condition: service_healthy
-    networks:
-      - app-network
-
-  nginx:
-    image: nginx:alpine
-    container_name: ai-nginx
-    ports:
-      - "80:80"
-      - "443:443"
-    volumes:
-      - ./nginx/docker-compose.conf:/etc/nginx/conf.d/default.conf:ro
-    depends_on:
-      - backend
-      - frontend
-    networks:
-      - app-network
-
-networks:
-  app-network:
-    driver: bridge

docs/1-introduction-and-goals.md

@@ -4,7 +4,8 @@ Describes the relevant requirements and the driving forces that software archite
 
 ## Requirements Overview
 
-**Project name**: AI Allucanget Biz  
+**Project name**: All You Can GET AI
+**URL**: [https://ai.allucanget.biz](https://ai.allucanget.biz)
 **Purpose**: Provide AI‑powered text, image, and video generation services via a web application.
 
 Users can choose between different AI models for:
@@ -14,6 +15,8 @@ Users can choose between different AI models for:
 - Text‑to‑video generation
 - Image‑to‑video generation
 
+Users can create accounts, log in, and view their generation history in a gallery. An admin dashboard allows managing users, models, and video generation jobs.
+
 ## Quality Goals
 
 | Priority | Quality Goal      | Scenario                                      |

docs/2-constraints.md

@@ -22,5 +22,5 @@ Any requirement that constrains software architects in their freedom of design a
 
 | Convention           | Background / Motivation                             |
 | -------------------- | --------------------------------------------------- |
-| Python 3.11+         | Modern language features, type hints                |
+| Python 3.12+         | Modern language features, type hints                |
 | pytest for all tests | Consistent test tooling across backend and frontend |

docs/5-building-block-view.md

@@ -5,21 +5,21 @@ Static decomposition of the system into building blocks (modules, components, su
 ## Level 1 – Whitebox Overall System
 
 ```text
-┌───────────────────────┐
+┌────────────────────────┐
 │  Frontend (Flask)      │
-└───────┬───────────────┘
+└───────┬────────────────┘
         │ REST API calls
-┌───────▼───────────────┐
+┌───────▼────────────────┐
 │  FastAPI Backend       │
 │  ├─ Auth Service       │
 │  ├─ User Service       │
 │  ├─ AI Service         │
 │  └─ DB Service (DuckDB)│
-└───────┬───────────────┘
+└───────┬────────────────┘
         │ DB access
-┌───────▼───────────────┐
+┌───────▼────────────────┐
 │  DuckDB Database       │
-└───────────────────────┘
+└────────────────────────┘
 ```
 
 **Motivation:** Separating the UI (Flask) from the API (FastAPI) allows independent scaling and testing of each layer.
@@ -66,17 +66,25 @@ Self-service profile management and admin user CRUD.
 Operational endpoints for application management.
 
 | Method | Path                        | Auth required | Admin only | Description                                |
-| ------ | --------------------- | ------------- | ---------- | ------------------------------------- |
+| ------ | --------------------------- | ------------- | ---------- | ------------------------------------------ |
 | GET    | `/admin/stats`              | ✓             | ✓          | User counts by role, token activity        |
 | GET    | `/admin/health/db`          | ✓             | ✓          | DuckDB connectivity check                  |
 | POST   | `/admin/tokens/purge`       | ✓             | ✓          | Remove expired/revoked refresh tokens      |
+| GET    | `/admin/videos`             | ✓             | ✓          | List all video jobs with user emails       |
+| POST   | `/admin/videos/{id}/cancel` | ✓             | ✓          | Cancel a queued/processing video job       |
+| POST   | `/admin/videos/{id}/retry`  | ✓             | ✓          | Retry a failed/cancelled video job         |
+| DELETE | `/admin/videos/{id}`        | ✓             | ✓          | Permanently delete a video job             |
+| POST   | `/admin/videos/purge`       | ✓             | ✓          | Delete old completed/failed/cancelled jobs |
+| POST   | `/admin/videos/timed-out`   | ✓             | ✓          | Mark stale processing jobs as failed       |
+| GET    | `/admin/models`             | ✓             | ✓          | List cached OpenRouter models              |
+| POST   | `/admin/models/refresh`     | ✓             | ✓          | Refresh model cache from OpenRouter        |
 
 ### White Box AI Service (`/ai`, `/generate`)
 
 Model listing and multi-modal generation via openrouter.ai.
 
 | Method | Path                           | Auth required | Description                                                                                                         |
-| ------ | ---------------------------- | ------------- | ------------------------------------------------------------------------------------------------------------------- |
+| ------ | ------------------------------ | ------------- | ------------------------------------------------------------------------------------------------------------------- |
 | GET    | `/ai/models`                   | ✓             | List available OpenRouter models                                                                                    |
 | POST   | `/ai/chat`                     | ✓             | Multi-turn chat completion                                                                                          |
 | POST   | `/generate/text`               | ✓             | Single-prompt text generation (optional system prompt)                                                              |
@@ -84,10 +92,15 @@ Model listing and multi-modal generation via openrouter.ai.
 | POST   | `/generate/video`              | ✓             | Text-to-video (Sora 2 Pro, Veo 3.1 Fast) — returns `polling_url`                                                    |
 | POST   | `/generate/video/from-image`   | ✓             | Image-to-video — returns `polling_url`                                                                              |
 | GET    | `/generate/video/status`       | ✓             | Poll video generation status via `polling_url`                                                                      |
+| GET    | `/generate/images`             | ✓             | List current user's generated images                                                                                |
+| GET    | `/generate/images/{id}`        | ✓             | Get a single generated image                                                                                        |
+| GET    | `/generate/videos`             | ✓             | List current user's video jobs                                                                                      |
+| GET    | `/generate/videos/{id}`        | ✓             | Get a single video job                                                                                              |
+| POST   | `/generate/videos/{id}/cancel` | ✓             | Cancel a queued/processing video job                                                                                |
 
-**Video generation flow:** The `/generate/video` and `/generate/video/from-image` endpoints submit a job to OpenRouter's `/api/v1/videos` endpoint and return immediately with `status: "queued"` and a `polling_url`. Clients poll `/generate/video/status?polling_url=...` every 5 seconds until `status` is `"completed"` (returns `unsigned_urls`) or `"failed"`.
+**Video generation flow:** The `/generate/video` and `/generate/video/from-image` endpoints queue a job in the local database and return immediately with `status: "queued"`. A background worker (`video_worker.py`) submits the job to OpenRouter's `/api/v1/videos` endpoint, receives a `polling_url`, and polls it periodically until the job reaches `"completed"` or `"failed"`. The frontend polls `GET /generate/video/{id}/status` every 5 seconds to show live status updates.
 
-**Image generation routing:** The router auto-detects the model type — models containing `"flux"` or `"gpt-5-image-mini"` are routed to `/chat/completions` with `modalities: ["image"]`, while others (e.g. DALL-E 3) use the legacy `/images/generations` endpoint.
+**Image generation routing:** The router auto-detects the model type — models containing `"flux"` or `"gpt-5-image-mini"` are routed to `/chat/completions` with `modalities: ["image"]` (or `["image", "text"]` depending on cached output modalities), while others (e.g. DALL-E 3) use the legacy `/images/generations` endpoint.
 
 ### White Box DB Service (`db.py`)
 

docs/6-runtime-view.md

@@ -48,20 +48,31 @@ Describes concrete behavior and interactions of the system's building blocks in
 1. User submits video generation form with prompt, model, aspect ratio, resolution, and duration
 2. Flask POSTs to `POST /generate/video` with JWT header
 3. Auth Service validates JWT
-4. Backend calls OpenRouter `POST /api/v1/videos` with model, prompt, aspect_ratio, resolution, duration_seconds
-5. OpenRouter returns `{"id": "...", "polling_url": "..."}` with `status: "queued"`
-6. FastAPI returns `VideoResponse` with `polling_url` to Flask
-7. Flask renders result page with polling UI
-8. Frontend JavaScript polls `GET /generate/video/status?polling_url=...` every 5 seconds
-9. When `status` becomes `"completed"`, the response includes `unsigned_urls` — the video is displayed in a `<video>` element
-10. If `status` becomes `"failed"`, an error message is shown
+4. Backend inserts a row into `generated_videos` with `status: "queued"` and returns the DB job ID
+5. Flask renders result page with polling UI
+6. Background worker (`video_worker.py`) picks up queued jobs every 15 seconds:
+   - Calls OpenRouter `POST /api/v1/videos` with model, prompt, and parameters
+   - Receives `{"id": "...", "polling_url": "..."}` and updates the DB row to `status: "processing"`
+   - Polls the `polling_url` every 15 seconds until `status` is `"completed"` or `"failed"`
+   - Updates the DB row with the final status and video URL
+7. Frontend JavaScript polls `GET /generate/video/{db_id}/status` every 5 seconds
+8. When `status` becomes `"completed"`, the response includes `video_url` — the video is displayed in a `<video>` element
+9. If `status` becomes `"failed"`, an error message is shown
+10. User can click "Cancel Job" to mark the job as `"cancelled"` (stops local polling, does not stop the provider job)
 
 ## Scenario 4a: Video Generation (Image-to-Video)
 
 1. User provides an image URL, motion prompt, model, aspect ratio, resolution, and duration
 2. Flask POSTs to `POST /generate/video/from-image` with JWT header
-3. Backend calls OpenRouter `POST /api/v1/videos` with `image_url`, prompt, and parameters
-4. Same polling flow as Scenario 4
+3. Same background worker flow as Scenario 4, with `generation_type: "image_to_video"`
+
+## Scenario 4b: Video Job Cancellation
+
+1. User clicks "Cancel Job" on the video detail page or gallery pending card
+2. Frontend POSTs to `/generate/video/{id}/cancel`
+3. Backend verifies the job belongs to the user and is not in a terminal state
+4. Backend updates the DB row `status` to `"cancelled"`
+5. Frontend stops polling and updates the UI to show "Job cancelled"
 
 ## Scenario 5: Token Refresh
 

docs/7-deployment-view.md

@@ -5,102 +5,79 @@ Describes:
 1. Technical infrastructure used to execute your system, with infrastructure elements like geographical locations, environments, computers, processors, channels and net topologies.
 2. Mapping of (software) building blocks to that infrastructure elements.
 
+**See**: [Coolify Deployment Guide](./deployment/coolify.md) for detailed instructions.
+
 ## Infrastructure Level 1
 
-```text
-┌────────────────────────────────────────────┐
-│  Host / VM                                 │
-│  ┌─────────────┐  ┌────────────────────┐   │
-│  │  frontend   │  │      backend       │   │
-│  │  (Flask)    │  │    (FastAPI)       │   │
-│  │  :12016     │  │    :12015          │   │
-│  └──────┬──────┘  └─────────┬──────────┘   │
-│         │                   │              │
-│         └────────┬──────────┘              │
-│                  │                         │
-│          ┌───────▼────────┐                │
-│          │  db (DuckDB)   │                │
-│          │  data/app.db   │                │
-│          └────────────────┘                │
-└────────────────────────────────────────────┘
+Hosted on a single VM running docker containers, deployed via Coolify with Nixpacks to 192.168.88.18 for production.
+
+Containers run behind nginx at 192.168.88.11 which handles TLS termination and reverse proxying to the frontend on port 12016 and backend on port 12015. The database is a file on the host filesystem at `data/app.db` accessed by the backend service.
+
+```mermaid
+graph TD
+      Users[Users / Internet]
+      Nginx[nginx reverse proxy\nTLS termination]
+      Users -->|HTTPS| Nginx
+
+      subgraph Coolify Server
+         direction TB
+         subgraph AI Frontend
+            AI_Frontend[AI Frontend\nFlask\nServes HTML/CSS/JS UI]
+         end
+         subgraph AI Backend
+            AI_Backend[AI Backend\nFastAPI\nCommunicates with openrouter.ai API]
+            db[(DuckDB Database\nFile: data/app.db)]
+            AI_Backend --> db
+         end
+         AI_Frontend -->|BACKEND_URL:12015| AI_Backend
+      end
+      Nginx -->|12016| AI_Frontend
 ```
 
-**Motivation:** All three components run on a single VM (or as Docker containers) for simplicity and low operational overhead.
+**Motivation:** All three components run as Docker containers for simplicity and low operational overhead.
 
 **Quality and/or Performance Features:** The frontend and backend are stateless; DuckDB persists data on the host filesystem.
 
 **Mapping of Building Blocks to Infrastructure:**
 
 | Building Block  | Container / Process          | Port            |
-| --------------- | ---------------------------- | ----- |
+| --------------- | ---------------------------- | --------------- |
+| Nginx           | `nginx`                      | 80/443 (public) |
+| Coolify Server  | `coolify`                    | —               |
 | Flask frontend  | `frontend`                   | 12016           |
 | FastAPI backend | `backend`                    | 12015           |
 | DuckDB          | File on host (`data/app.db`) | —               |
 
 ## Infrastructure Level 2
 
-### Docker Compose (Recommended for Development & Production)
+### Coolify with Nixpacks (Production)
 
-All services are containerized and orchestrated with `docker compose`:
+Both services are deployed as separate Nixpacks resources in Coolify, which results in two separate containers running on the same host. The database is a file on the host filesystem, mounted as a volume in the backend container.
 
-```text
-┌─────────────────────────────────────────────────────────────┐
-│  Docker Host / VM                                           │
-│  ┌──────────────────────────────────────────────────────┐   │
-│  │  Docker Network: app-network (bridge)                │   │
-│  │  ┌──────────────────────────────────────────────┐    │   │
-│  │  │ Backend Container (FastAPI)                  │    │   │
-│  │  │ - Port: 12015                                │    │   │
-│  │  │ - Service Name: backend                      │    │   │
-│  │  │ - Volume Mount: /app/data ← host/data/       │    │   │
-│  │  ├──────────────────────────────────────────────┤    │   │
-│  │  │ Frontend Container (Flask)                   │    │   │
-│  │  │ - Port: 12016                                │    │   │
-│  │  │ - Service Name: frontend                     │    │   │
-│  │  │ - Depends on: backend (health check)         │    │   │
-│  │  ├──────────────────────────────────────────────┤    │   │
-│  │  │ Nginx Container (Reverse Proxy)              │    │   │
-│  │  │ - Port: 80 (HTTP), 443 (HTTPS)               │    │   │
-│  │  │ - Config: nginx/docker-compose.conf          │    │   │
-│  │  │ - Routes: /api/* → backend:12015             │    │   │
-│  │  │           / → frontend:12016                 │    │   │
-│  │  └──────────────────────────────────────────────┘    │   │
-│  └──────────────────────────────────────────────────────┘   │
-│                         ▲                                   │
-│                  Host Port Bindings                         │
-│         80:80, 443:443, 12015:12015, 12016:12016            │
-└─────────────────────────────────────────────────────────────┘
-                         │
-                      Users / Internet
+#### Frontend
+
+```mermaid
+graph TD
+   subgraph Coolify Server
+      direction TB
+      subgraph AI Frontend
+         AI_Frontend[AI Frontend\nNixpacks\nBase Dir: /frontend]
+      end
+   end
+   Users[Users / Internet] -->|HTTPS| AI_Frontend
 ```
 
-**Deployment Steps:**
+#### Backend
 
-1. Ensure Docker and Docker Compose are installed
-2. Create `.env` with required environment variables
-3. Run: `docker compose up --build`
-4. Access via browser at `http://localhost:12016` or through Nginx at `http://localhost:80`
-
-**Benefits:**
-
-- **Consistency**: Same containerized environment across development, testing, and production
-- **Simplicity**: Single command to start entire stack
-- **Portability**: Run on any system with Docker installed
-- **Persistence**: DuckDB data survives container restarts via volume mounts
-- **Networking**: Service names enable automatic DNS resolution (backend:12015, frontend:12016)
-- **Observability**: Easy logging with `docker compose logs`
-
-**See**: [Docker Compose Deployment Guide](./deployment/docker-compose.md) for detailed instructions.
-
-### Coolify (Alternative for Existing Deployments)
-
-If using Coolify instead of Docker Compose:
-
-1. **Recommended Path**: Deploy `docker-compose.coolify.yml` as one Coolify Docker Compose resource
-2. **Public Entry Point**: Route domain to `nginx` service on port 80; backend and frontend stay internal
-3. **Data Persistence**: Named volume keeps DuckDB data at `/app/data`
-4. **Fallback Path**: Use separate Nixpacks services only when one-stack Compose deploy is not suitable
-
-**Note**: In Compose-based Coolify deployment, frontend reaches backend through Docker DNS at `http://backend:12015`. In Nixpacks-based deployment, use Coolify internal networking or colocate both services.
-
-**See**: [Coolify Deployment Guide](./deployment/coolify.md) for detailed instructions.
+```mermaid
+graph TD
+   subgraph Coolify Server
+      direction TB
+      subgraph AI Backend
+         AI_Backend[AI Backend\nNixpacks\nBase Dir: /backend]
+         db[(DuckDB Database\nVolume: /app/data)]
+         AI_Backend --> db
+      end
+   end
+   Frontend[Frontend Container] -->|BACKEND_URL:12015| AI_Backend
+```

docs/8-crosscutting-concepts.md

@@ -4,6 +4,14 @@ Describes crosscutting concepts (practices, patterns, regulations or solution id
 
 > Pick **only** the most-needed topics for your system.
 
+## OpenRouter API Integration
+
+see [docs/8.1-openrouter.md](./8.1-openrouter.md) for details on how the backend integrates with OpenRouter for multi-modal AI generation, including image and video generation flows.
+
+## DuckDB Concurrency and Storage
+
+See [docs/8.2-duckdb.md](./8.2-duckdb.md) for details on how the backend handles concurrent access to DuckDB and manages the database file on the host filesystem.
+
 ## Security
 
 - All API endpoints (except `/auth/login`) require a valid JWT in the `Authorization: Bearer` header.
@@ -25,72 +33,3 @@ Describes crosscutting concepts (practices, patterns, regulations or solution id
 
 - All secrets (API keys, DB path, JWT secret) loaded from environment variables or `.env` file.
 - No secrets committed to source control.
-
-## DuckDB Concurrency and Storage
-
-### Single Writer Per Process
-
-DuckDB allows only one process to open the database file in read-write mode at a time. The FastAPI backend must be run with a single worker (`uvicorn --workers 1`). Running multiple workers against the same DuckDB file will cause startup errors.
-
-### asyncio.Lock for Writes
-
-All database write operations (`INSERT`, `UPDATE`, `DELETE`) in the FastAPI async context are wrapped in a single `asyncio.Lock` (`get_write_lock()` from `backend/app/db.py`). This prevents concurrent coroutines from issuing overlapping writes within the single process, which would otherwise raise DuckDB optimistic concurrency errors.
-
-Read operations (`SELECT`) do not require the lock — DuckDB's MVCC provides consistent read snapshots.
-
-### Schema
-
-```sql
-CREATE TABLE users (
-    id            UUID      DEFAULT uuid() PRIMARY KEY,
-    email         VARCHAR   NOT NULL UNIQUE,
-    password_hash VARCHAR   NOT NULL,
-    role          VARCHAR   DEFAULT 'user',
-    created_at    TIMESTAMP DEFAULT now(),
-    updated_at    TIMESTAMP DEFAULT now()
-);
-
-CREATE TABLE refresh_tokens (
-    jti        UUID      DEFAULT uuid() PRIMARY KEY,
-    user_id    UUID      NOT NULL,           -- soft FK to users.id
-    issued_at  TIMESTAMP DEFAULT now(),
-    expires_at TIMESTAMP NOT NULL,
-    revoked    BOOLEAN   DEFAULT false
-);
-```
-
-> The `REFERENCES users(id)` foreign key is intentionally omitted from `refresh_tokens`. DuckDB fires FK checks on `UPDATE` of the parent table (including email changes), causing false constraint violations. Referential integrity is enforced manually: deleting a user also deletes their refresh tokens in the same write transaction.
-
-### Access Tokens
-
-Access tokens are **stateless** JWTs — not stored in the database. They are validated by signature and expiry claim only. The short TTL (15 minutes) limits the blast radius if a token is leaked.
-
-### Refresh Tokens
-
-Refresh tokens store a JTI (JWT ID) UUID in the `refresh_tokens` table. On each use the old JTI is revoked and a new one issued (rotation). On logout the JTI is immediately revoked. Expired and revoked tokens can be purged via `POST /admin/tokens/purge`.
-
-### Future: AI Generation History
-
-AI generation metadata (model, prompt, cost, result URLs) can be stored as JSON columns in a future `generation_history` table in DuckDB, enabling per-user analytics and usage dashboards at zero extra infrastructure cost.
-
-## OpenRouter API Integration
-
-### Image Generation
-
-Image generation uses two different OpenRouter endpoints depending on the model:
-
-- **Legacy endpoint** (`/images/generations`): Used by DALL-E 3 and similar models. Returns `data[].url` and `data[].b64_json`.
-- **Chat completions** (`/chat/completions` with `modalities: ["image"]`): Used by FLUX.2 Klein 4B and GPT-5 Image Mini. Returns `choices[0].message.images[].image_url.url` as base64 data URLs.
-
-The router auto-detects the model type and routes accordingly. Image configuration (`aspect_ratio`, `image_size`) is passed via `image_config` for chat-based models.
-
-### Video Generation
-
-Video generation uses OpenRouter's `/api/v1/videos` endpoint with a **submit-and-poll** pattern:
-
-1. `POST /api/v1/videos` with `model`, `prompt`, `aspect_ratio`, `resolution`, `duration_seconds`
-2. Response: `{"id": "job_id", "polling_url": "https://..."}` with `status: "queued"`
-3. Poll `GET polling_url` every 5 seconds until `status` is `"completed"` or `"failed"`
-4. Completed response includes `unsigned_urls: [str]` array with video download URLs
-
-Supported models: `openai/sora-2-pro`, `google/veo-3.1-fast`. Both text-to-video and image-to-video use the same `/api/v1/videos` endpoint (image-to-video includes `image_url` in the request body).

docs/8.1-openrouter.md

@@ -0,0 +1,31 @@
+# OpenRouter API Integration
+
+## Text Generation
+
+> [!warning]
+> TODO: Add more details on how the backend integrates with OpenRouter for text generation, including chat completions and single-prompt generation flows.
+
+## Image Generation
+
+Image generation uses two different OpenRouter endpoints depending on the model:
+
+- **Legacy endpoint** (`/images/generations`): Used by DALL-E 3 and similar models. Returns `data[].url` and `data[].b64_json`.
+- **Chat completions** (`/chat/completions` with `modalities: ["image"]`): Used by FLUX.2 Klein 4B and GPT-5 Image Mini. Returns `choices[0].message.images[].image_url.url` as base64 data URLs.
+
+The router auto-detects the model type and routes accordingly. Image configuration (`aspect_ratio`, `image_size`) is passed via `image_config` for chat-based models.
+
+## Video Generation
+
+Video generation uses OpenRouter's `/api/v1/videos` endpoint with a **submit-and-poll** pattern orchestrated by a background worker:
+
+1. User submits a video request via `POST /generate/video` (or `/generate/video/from-image`)
+2. Backend inserts a row into `generated_videos` with `status: "queued"` and returns immediately
+3. Background worker (`video_worker.py`) picks up queued jobs every 15 seconds:
+   - Calls `POST /api/v1/videos` with `model`, `prompt`, `aspect_ratio`, `resolution`, `duration`
+   - Receives `{"id": "job_id", "polling_url": "https://..."}` and updates DB to `status: "processing"`
+   - Polls `GET polling_url` every 15 seconds until `status` is `"completed"` or `"failed"`
+   - Updates DB with final status, `video_url`, and any `error` message
+4. Frontend polls `GET /generate/video/{db_id}/status` every 5 seconds to show live updates
+5. Completed response includes `video_url` — the video is displayed in a `<video>` element
+
+Supported models: `openai/sora-2-pro`, `google/veo-3.1-fast`. Both text-to-video and image-to-video use the same `/api/v1/videos` endpoint (image-to-video includes `frame_images` with `first_frame` in the request body).

docs/8.2-database.md

@@ -0,0 +1,46 @@
+# DuckDB Concurrency and Storage
+
+## Single Writer Per Process
+
+DuckDB allows only one process to open the database file in read-write mode at a time. The FastAPI backend must be run with a single worker (`uvicorn --workers 1`). Running multiple workers against the same DuckDB file will cause startup errors.
+
+## asyncio.Lock for Writes
+
+All database write operations (`INSERT`, `UPDATE`, `DELETE`) in the FastAPI async context are wrapped in a single `asyncio.Lock` (`get_write_lock()` from `backend/app/db.py`). This prevents concurrent coroutines from issuing overlapping writes within the single process, which would otherwise raise DuckDB optimistic concurrency errors.
+
+Read operations (`SELECT`) do not require the lock — DuckDB's MVCC provides consistent read snapshots.
+
+## Schema
+
+```sql
+CREATE TABLE users (
+    id            UUID      DEFAULT uuid() PRIMARY KEY,
+    email         VARCHAR   NOT NULL UNIQUE,
+    password_hash VARCHAR   NOT NULL,
+    role          VARCHAR   DEFAULT 'user',
+    created_at    TIMESTAMP DEFAULT now(),
+    updated_at    TIMESTAMP DEFAULT now()
+);
+
+CREATE TABLE refresh_tokens (
+    jti        UUID      DEFAULT uuid() PRIMARY KEY,
+    user_id    UUID      NOT NULL,           -- soft FK to users.id
+    issued_at  TIMESTAMP DEFAULT now(),
+    expires_at TIMESTAMP NOT NULL,
+    revoked    BOOLEAN   DEFAULT false
+);
+```
+
+> The `REFERENCES users(id)` foreign key is intentionally omitted from `refresh_tokens`. DuckDB fires FK checks on `UPDATE` of the parent table (including email changes), causing false constraint violations. Referential integrity is enforced manually: deleting a user also deletes their refresh tokens in the same write transaction.
+
+## Access Tokens
+
+Access tokens are **stateless** JWTs — not stored in the database. They are validated by signature and expiry claim only. The short TTL (15 minutes) limits the blast radius if a token is leaked.
+
+## Refresh Tokens
+
+Refresh tokens store a JTI (JWT ID) UUID in the `refresh_tokens` table. On each use the old JTI is revoked and a new one issued (rotation). On logout the JTI is immediately revoked. Expired and revoked tokens can be purged via `POST /admin/tokens/purge`.
+
+## Future: AI Generation History
+
+AI generation metadata (model, prompt, cost, result URLs) can be stored as JSON columns in a future `generation_history` table in DuckDB, enabling per-user analytics and usage dashboards at zero extra infrastructure cost.

docs/ARCHITECTURE.md

@@ -1,6 +1,6 @@
 # Architecture Documentation
 
-This file is the entry point for the architecture documentation of **AI Allucanget Biz**.
+This file is the entry point for the architecture documentation of **All You Can GET AI Biz**.
 
 The documentation follows the [arc42 template](https://arc42.org/overview) and is split into 12 section files, each covering a specific aspect of the architecture. Read the sections in order for a full picture, or jump directly to the section most relevant to you.
 

docs/deployment/coolify.md

@@ -1,30 +1,6 @@
 # Coolify Deployment Guide
 
-This guide covers deploying `ai.allucanget.biz` using [Coolify](https://coolify.io) from the repository `https://git.allucanget.biz/allucanget/ai.allucanget.biz.git`.
-
-## Deployment Options
-
-The application supports two deployment approaches:
-
-### Option 1: Docker Compose (Recommended)
-
-- **Best for**: Coolify deployments, local development, consistent environments
-- **Compose file**: `docker-compose.coolify.yml` in Coolify, `docker-compose.yml` locally
-- **Key benefits**:
-  - Same containerized services locally and in production
-  - Built-in service orchestration and networking
-  - Easy volume management for data persistence
-  - Better debugging and log viewing
-
-### Option 2: Coolify with Nixpacks (Alternative)
-
-- **Best for**: Existing Coolify installations, serverless deployments
-- **Key differences**:
-  - Uses Coolify's Nixpacks build system instead of Docker
-  - Requires separate services setup in Coolify UI
-  - Manual environment variable configuration per service
-
-This guide covers both Coolify paths. Use **Option 1 (Docker Compose)** for new deployments. Use **Option 2 (Nixpacks)** only if you need separate Coolify services instead of one stack.
+This guide covers deploying `ai.allucanget.biz` using [Coolify](https://coolify.io) with Nixpacks from the repository `https://git.allucanget.biz/allucanget/ai.allucanget.biz.git`.
 
 ## Architecture Overview
 
@@ -46,55 +22,6 @@ Coolify's built-in reverse proxy routes traffic:
 - Git repository pushed to `https://git.allucanget.biz/allucanget/ai.allucanget.biz.git`
 - Domain configured to point to your Coolify server
 
-## Option 1: Deploy with Docker Compose in Coolify
-
-Use Coolify's **Docker Compose** resource with the dedicated compose file `docker-compose.coolify.yml`.
-
-### Step 1: Create Docker Compose Resource
-
-1. In Coolify, click **Add Resource** → **Deploy a new resource** → **Docker Compose**
-2. Connect your Git repository (`git.allucanget.biz`)
-3. Select the `ai.allucanget.biz` repository and `main` branch
-4. Set **Compose File** to `docker-compose.coolify.yml`
-5. Set **Base Directory** to `/`
-6. Select the `frontend` service as the public-facing service on port `12016`
-7. Click **Create Resource**
-
-### Step 2: Configure Environment Variables
-
-Add these in Coolify before first deploy:
-
-| Variable             | Service    | Example                     |
-| -------------------- | ---------- | --------------------------- |
-| `OPENROUTER_API_KEY` | `backend`  | `sk-or-v1-...`              |
-| `JWT_SECRET`         | `backend`  | `openssl rand -hex 32`      |
-| `APP_URL`            | `backend`  | `https://ai.allucanget.biz` |
-| `APP_NAME`           | `backend`  | `AI Allucanget`             |
-| `CORS_ORIGINS`       | `backend`  | `https://ai.allucanget.biz` |
-| `FLASK_SECRET_KEY`   | `frontend` | `openssl rand -hex 32`      |
-| `BACKEND_URL`        | `frontend` | `http://backend:12015`      |
-
-### Step 3: Configure Domain and Proxy
-
-1. Attach domain `ai.allucanget.biz` to the `frontend` service on port `12016`
-2. Enable **Auto HTTPS** in Coolify
-3. Coolify terminates TLS and forwards traffic directly to `frontend:12016`
-
-No nginx service is needed in the Coolify stack. The frontend Flask app proxies `/api/*` calls to `http://backend:12015` internally. Coolify's own reverse proxy handles external TLS termination.
-
-### Step 4: Persistent Storage
-
-`docker-compose.coolify.yml` uses named volume `app-data` for DuckDB persistence at `/app/data`. No extra host bind mount needed for default setup.
-
-### Step 5: Deploy and Verify
-
-1. Start deployment in Coolify
-2. Confirm `backend` becomes healthy
-3. Open `https://ai.allucanget.biz/health` or proxied health route configured in Coolify
-4. Open main domain and verify frontend loads
-
-## Option 2: Deploy with Nixpacks in Coolify
-
 ## Step 1: Create Backend Service
 
 1. In Coolify, click **Add Resource** → **Deploy a new resource** → **Git**
@@ -102,7 +29,7 @@ No nginx service is needed in the Coolify stack. The frontend Flask app proxies
 3. Select the `ai.allucanget.biz` repository
 4. Choose the `main` branch
 5. Set **Build Pack** to `nixpacks`
-6. **CRITICAL: Set Base Directory to `/backend`** — this tells Nixpacks to look in the `backend/` subdirectory for `requirements.txt` and the Python application
+6. Set **Base Directory** to `/backend` - this tells Nixpacks to look in the `backend/` subdirectory for `requirements.txt` and the Python application
 7. Set **Ports Exposed** to `12015`
 8. Set **Start Command** to:
 
@@ -123,7 +50,7 @@ Add these as **Runtime** environment variables in Coolify:
 | `OPENROUTER_API_KEY` | OpenRouter API key for AI generation | `sk-or-v1-...`                       |
 | `JWT_SECRET`         | Secret key for JWT token signing     | Generate with `openssl rand -hex 32` |
 | `APP_URL`            | Public URL of the backend            | `https://api.ai.allucanget.biz`      |
-| `APP_NAME`           | Application name                     | `AI Allucanget`                      |
+| `APP_NAME`           | Application name                     | `All You Can GET AI`                 |
 | `CORS_ORIGINS`       | Comma-separated allowed origins      | `https://ai.allucanget.biz`          |
 
 ## Step 2: Create Frontend Service
@@ -132,7 +59,7 @@ Add these as **Runtime** environment variables in Coolify:
 2. Select the same repository
 3. Choose the `main` branch
 4. Set **Build Pack** to `nixpacks`
-5. **CRITICAL: Set Base Directory to `/frontend`** — this tells Nixpacks to look in the `frontend/` subdirectory for `requirements.txt` and the Python application
+5. Set **Base Directory** to `/frontend` - this tells Nixpacks to look in the `frontend/` subdirectory for `requirements.txt` and the Python application
 6. Set **Ports Exposed** to `12016`
 7. Set **Start Command** to:
 
@@ -142,7 +69,7 @@ Add these as **Runtime** environment variables in Coolify:
 
 8. Click **Create Resource**
 
-> **Note:** The frontend uses `requirements.txt` for production dependencies and `requirements-dev.txt` for development dependencies (like pytest). Nixpacks will automatically detect and install only the production dependencies.
+> **Note:** Nixpacks will automatically detect and install only the production dependencies from `requirements.txt`.
 > **Important:** Nixpacks copies the **contents** of the Base Directory to `/app/` in the container. When Base Directory is `/frontend`, the `frontend/` folder wrapper is removed — only `app/`, `tests/`, and `requirements.txt` are copied. Therefore the start command uses `app.main:app` (not `frontend.app.main:app`).
 
 ### Frontend Environment Variables
@@ -170,33 +97,6 @@ Coolify provides a built-in reverse proxy. Configure routing rules:
 - **Port**: `12016`
 - **Path**: `/` → forward to frontend
 
-### Nginx Configuration (Optional)
-
-If you need custom Nginx configuration, create `nginx/coolify.conf`:
-
-```nginx
-# Reverse proxy configuration for Coolify
-# This file is for reference — Coolify's built-in proxy handles routing
-
-# Backend API proxy
-location /api/ {
-    proxy_pass http://backend:12015;
-    proxy_set_header Host $host;
-    proxy_set_header X-Real-IP $remote_addr;
-    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-    proxy_set_header X-Forwarded-Proto $scheme;
-}
-
-# Frontend proxy
-location / {
-    proxy_pass http://frontend:12016;
-    proxy_set_header Host $host;
-    proxy_set_header X-Real-IP $remote_addr;
-    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-    proxy_set_header X-Forwarded-Proto $scheme;
-}
-```
-
 ## Step 4: SSL/TLS
 
 Enable HTTPS in Coolify for both services:
@@ -219,12 +119,6 @@ If you want to persist DuckDB data:
 
 ## Troubleshooting
 
-### Docker Compose deployment fails in Coolify
-
-- Verify Coolify uses `docker-compose.coolify.yml`, not local `docker-compose.yml`
-- Verify public domain points to `frontend` service on port `12016`
-- Do not add `nginx` to the Coolify stack — bind-mounting a local config file will fail since the file doesn't exist on the Coolify server
-
 ### Backend healthcheck stays unhealthy
 
 - Check backend logs in Coolify
@@ -259,11 +153,11 @@ If you want to persist DuckDB data:
 All required environment variables:
 
 | Variable             | Service  | Required                              |
-| -------------------- | -------- | -------------------------------- |
+| -------------------- | -------- | ------------------------------------- |
 | `OPENROUTER_API_KEY` | Backend  | Yes                                   |
 | `JWT_SECRET`         | Backend  | Yes                                   |
 | `APP_URL`            | Backend  | Yes                                   |
-| `APP_NAME`           | Backend  | No (defaults to "AI Allucanget") |
+| `APP_NAME`           | Backend  | No (defaults to "All You Can GET AI") |
 | `CORS_ORIGINS`       | Backend  | Yes                                   |
 | `FLASK_SECRET_KEY`   | Frontend | Yes                                   |
 | `BACKEND_URL`        | Frontend | Yes                                   |
@@ -271,8 +165,6 @@ All required environment variables:
 ## Deployment Checklist
 
 - [ ] Repository pushed to Git
-- [ ] For Docker Compose: Coolify resource uses `docker-compose.coolify.yml`
-- [ ] For Docker Compose: domain points to `frontend` service on port `12016`
 - [ ] Backend service created with correct base directory (`/backend`)
 - [ ] Backend environment variables configured
 - [ ] Frontend service created with correct base directory (`/frontend`)

docs/deployment/docker-compose.md

@@ -1,243 +0,0 @@
-# Docker Compose Deployment Guide
-
-This guide explains how to deploy the AI application locally using Docker Compose.
-
-## Overview
-
-Docker Compose provides a convenient way to:
-
-- **Develop locally** with the same containerized environment used in production
-- **Manage services** as a single stack (backend, frontend, and Nginx)
-- **Persist data** using volume mounts
-- **Enable service-to-service communication** using Docker's internal networking
-
-## Prerequisites
-
-- Docker Desktop installed and running
-- Docker Compose v2.0+ (included with Docker Desktop)
-- A `.env` file with required environment variables (see [Configuration](#configuration))
-
-## Quick Start
-
-1. **Clone the repository** and navigate to the project root:
-
-```bash
-cd ai.allucanget.biz
-```
-
-2. **Configure environment variables** in `.env`:
-
-```bash
-# Copy the example if you don't have a .env file
-cp .env.example .env
-
-# Edit .env and fill in required values
-# - OPENROUTER_API_KEY: Your openrouter.ai API key
-# - JWT_SECRET: A random secret for signing JWT tokens
-# - FLASK_SECRET_KEY: A random secret for Flask session cookies
-```
-
-3. **Build and start the services**:
-
-```bash
-docker compose up --build
-```
-
-4. **Access the application**:
-   - Frontend UI: http://localhost:12016
-   - Backend API: http://localhost:12015
-   - API Health: http://localhost:12015/health
-   - Nginx Proxy: http://localhost:80 (routes to frontend) or http://localhost:80/api (routes to backend)
-
-5. **View logs**:
-
-```bash
-# All services
-docker compose logs -f
-
-# Specific service
-docker compose logs -f backend
-docker compose logs -f frontend
-```
-
-6. **Stop the services**:
-
-```bash
-docker compose down
-```
-
-## Configuration
-
-### Environment Variables
-
-The `.env` file in the project root is automatically loaded by Docker Compose. Required variables:
-
-| Variable             | Purpose                                    | Example                                                                      |
-| -------------------- | ------------------------------------------ | ---------------------------------------------------------------------------- |
-| `OPENROUTER_API_KEY` | API key for AI generation                  | `sk-or-v1-...`                                                               |
-| `JWT_SECRET`         | Secret for JWT token signing               | (random hex string)                                                          |
-| `FLASK_SECRET_KEY`   | Secret for Flask session cookies           | (random hex string)                                                          |
-| `BACKEND_URL`        | Internal URL for frontend to reach backend | `http://backend:12015` (Docker) or `http://localhost:12015` (local dev)      |
-| `CORS_ORIGINS`       | Allowed CORS origins for backend           | `http://localhost:12016` (local) or `https://ai.allucanget.biz` (production) |
-| `APP_URL`            | Public URL of the backend                  | `http://localhost` or `https://ai.allucanget.biz`                            |
-| `APP_NAME`           | Application name                           | `AI Allucanget`                                                              |
-
-### Volume Mounts
-
-The `docker-compose.yml` defines the following volume:
-
-- `./data:/app/data` — Persists the DuckDB database (`app.db`) between container restarts
-
-## Architecture
-
-```
-┌─────────────────────────────────────────────────────────┐
-│ Docker Network: app-network (internal bridge)          │
-├─────────────────────────────────────────────────────────┤
-│                                                         │
-│  ┌──────────────────┐    ┌──────────────────────────┐  │
-│  │  Frontend        │    │  Backend                 │  │
-│  │  Port 12016      │    │  Port 12015              │  │
-│  │  Flask + Gunicorn│◄──►│  FastAPI + Uvicorn       │  │
-│  │  (Service name:  │    │  (Service name: backend) │  │
-│  │   frontend)      │    │                          │  │
-│  └──────────────────┘    │  ┌────────────────────┐  │  │
-│       ▲                  │  │ DuckDB Database    │  │  │
-│       │                  │  │ /app/data/app.db   │  │  │
-│  ┌────▼──────────────┐   │  └────────────────────┘  │  │
-│  │  Nginx (Port 80)  │   └──────────────────────────┘  │
-│  │  Reverse Proxy    │                                 │
-│  └───────────────────┘                                 │
-│       ▲                                                │
-│       │ (bind mount)                                   │
-│   Host:80, 443                                         │
-└─────────────────────────────────────────────────────────┘
-                       │
-                   Host OS
-```
-
-## Common Tasks
-
-### View service status
-
-```bash
-docker compose ps
-```
-
-Output example:
-
-```
-NAME                COMMAND                  SERVICE      STATUS      PORTS
-ai-backend          "uvicorn app.main:a…"   backend      Up 2m       0.0.0.0:12015->12015/tcp
-ai-frontend         "gunicorn app.main:…"   frontend     Up 2m       0.0.0.0:12016->12016/tcp
-ai-nginx            "nginx -g daemon of…"   nginx        Up 2m       0.0.0.0:80->80/tcp, 0.0.0.0:443->443/tcp
-```
-
-### Rebuild images
-
-```bash
-# Rebuild all services
-docker compose build
-
-# Rebuild a specific service
-docker compose build backend
-docker compose build frontend
-```
-
-### Access container shell
-
-```bash
-# Backend shell
-docker compose exec backend bash
-
-# Frontend shell
-docker compose exec frontend bash
-```
-
-### View service logs
-
-```bash
-# Tail all logs
-docker compose logs -f
-
-# Follow only backend logs
-docker compose logs -f backend
-
-# Last 100 lines of frontend logs
-docker compose logs --tail 100 frontend
-```
-
-### Clean up
-
-```bash
-# Stop containers but keep volumes and images
-docker compose stop
-
-# Stop and remove containers
-docker compose down
-
-# Stop containers and remove volumes (WARNING: deletes database!)
-docker compose down -v
-
-# Remove images as well
-docker compose down --rmi all
-```
-
-## Troubleshooting
-
-### "Docker daemon is not running"
-
-**Error**: `error during connect: This error may indicate the client is not properly configured`
-
-**Solution**: Start Docker Desktop (Windows/Mac) or start the Docker daemon (Linux).
-
-### "Port 12015 or 12016 already in use"
-
-**Error**: `Error response from daemon: bind: address already in use`
-
-**Solution**: Either stop the other application using that port, or modify the ports in `docker-compose.yml`:
-
-```yaml
-ports:
-  - "13000:12015" # Maps host:13000 to container:12015
-  - "13001:12016" # Maps host:13001 to container:12016
-```
-
-### Frontend cannot reach backend
-
-**Error**: Connection refused or 502 Bad Gateway
-
-**Ensure**:
-
-1. Backend service is running: `docker compose ps`
-2. `BACKEND_URL` in `.env` is set to `http://backend:12015` (not `localhost`)
-3. Services are on the same Docker network: `docker compose ps` shows them in the same `docker-compose.yml`
-
-### Database file not persisting
-
-**Issue**: Data is lost after `docker compose down`
-
-**Solution**: The volume mount `./data:/app/data` should persist data automatically. Check that:
-
-1. The `data/` directory exists on your host
-2. The volume is mounted correctly in `docker-compose.yml`
-3. DuckDB has write permissions to the `data/` directory
-
-## Performance Tips
-
-- **Development**: Use `docker compose up` for real-time logs and quick iterations
-- **Background**: Use `docker compose up -d` to run services detached
-- **Rebuilds**: Use `docker compose up --build` only when dependencies or Dockerfiles change
-- **Caching**: Docker caches layers; unchanged steps build faster
-
-## Next Steps
-
-- **Deployment**: See [Coolify Deployment](./coolify.md) for production deployment
-- **Testing**: See [Testing Guide](../TESTING.md) for running tests
-- **Architecture**: See [Deployment View](../7-deployment-view.md) for system architecture
-
-## Additional Resources
-
-- [Docker Compose Documentation](https://docs.docker.com/compose/)
-- [Docker Networking Guide](https://docs.docker.com/network/)
-- [Best Practices for Writing Dockerfiles](https://docs.docker.com/develop/dev-best-practices/dockerfile_best-practices/)

frontend/app/main.py

@@ -1,9 +1,11 @@
 """Flask frontend application."""
 import functools
+from datetime import datetime, timezone
 
 import httpx
 from flask import (
     Flask,
+    Response,
     flash,
     jsonify,
     redirect,
@@ -23,6 +25,42 @@ app.config.from_object(Config)
 # Helpers
 # ---------------------------------------------------------------------------
 
+@app.template_filter("fromisoformat")
+def from_iso_format(s: str) -> datetime:
+    """Convert ISO 8601 string to datetime object."""
+    return datetime.fromisoformat(s)
+
+
+@app.template_filter("humantime")
+def human_time(dt: datetime) -> str:
+    """Format a datetime object into a human-readable relative time."""
+    now = datetime.now(timezone.utc)
+    # Ensure dt is aware for comparison
+    if dt.tzinfo is None:
+        dt = dt.replace(tzinfo=timezone.utc)
+
+    diff = now - dt
+    seconds = diff.total_seconds()
+
+    if seconds < 60:
+        return "just now"
+    elif seconds < 3600:
+        minutes = int(seconds / 60)
+        return f"{minutes} minute{'s' if minutes > 1 else ''} ago"
+    elif seconds < 86400:
+        hours = int(seconds / 3600)
+        return f"{hours} hour{'s' if hours > 1 else ''} ago"
+    elif seconds < 2592000:
+        days = int(seconds / 86400)
+        return f"{days} day{'s' if days > 1 else ''} ago"
+    elif seconds < 31536000:
+        months = int(seconds / 2592000)
+        return f"{months} month{'s' if months > 1 else ''} ago"
+    else:
+        years = int(seconds / 31536000)
+        return f"{years} year{'s' if years > 1 else ''} ago"
+
+
 def _backend(path: str) -> str:
     return f"{app.config['BACKEND_URL']}{path}"
 
@@ -34,6 +72,60 @@ def _api(method: str, path: str, *, token: str | None = None, **kwargs):
     return httpx.request(method, _backend(path), headers=headers, timeout=30, **kwargs)
 
 
+def _model_matches_modality(model: dict, modality: str) -> bool:
+    """Heuristic fallback when backend modality filter returns empty."""
+    model_modality = (model.get("modality") or "").lower()
+    if model_modality == modality:
+        return True
+
+    text = f"{model.get('id', '')} {model.get('name', '')}".lower()
+    keywords = {
+        "image": ["image", "dall-e", "flux", "stable-diffusion", "sdxl", "recraft", "ideogram", "gpt-image"],
+        "video": ["video", "sora", "runway", "veo", "kling", "pika", "luma", "wan"],
+        "audio": ["audio", "speech", "voice", "tts", "transcribe", "whisper"],
+    }
+
+    if modality in keywords:
+        return any(k in text for k in keywords[modality])
+
+    if modality == "text":
+        non_text_hits = any(
+            k in text for k in keywords["image"] + keywords["video"] + keywords["audio"])
+        return not non_text_hits
+
+    return False
+
+
+def _load_models(token: str, modality: str) -> list[dict]:
+    """Load models for modality; fallback to unfiltered cache if needed."""
+    try:
+        models_resp = _api("GET", "/models/", token=token,
+                           params={"modality": modality})
+    except httpx.RequestError:
+        return []
+    if models_resp.status_code == 200:
+        try:
+            models = models_resp.json()
+        except ValueError:
+            models = []
+        if models:
+            return models
+
+    try:
+        all_resp = _api("GET", "/models/", token=token)
+    except httpx.RequestError:
+        return []
+    if all_resp.status_code != 200:
+        return []
+
+    try:
+        all_models = all_resp.json()
+    except ValueError:
+        return []
+    filtered = [m for m in all_models if _model_matches_modality(m, modality)]
+    return filtered or all_models
+
+
 def login_required(view):
     @functools.wraps(view)
     def wrapped(*args, **kwargs):
@@ -121,11 +213,96 @@ def dashboard():
     token = session["access_token"]
     resp = _api("GET", "/users/me", token=token)
     user = resp.json() if resp.status_code == 200 else {}
-    return render_template("dashboard.html", user=user)
+    img_resp = _api("GET", "/images/", token=token)
+    images = img_resp.json() if img_resp.status_code == 200 else []
+    gen_resp = _api("GET", "/generate/images", token=token)
+    generated_images = gen_resp.json() if gen_resp.status_code == 200 else []
+
+    vid_resp = _api("GET", "/generate/videos", token=token)
+    videos = vid_resp.json() if vid_resp.status_code == 200 else []
+    pending_videos = [v for v in videos if v.get(
+        "status") not in ("completed", "failed")]
+    completed_videos = [v for v in videos if v.get("status") == "completed"]
+
+    return render_template("dashboard.html", user=user, images=images,
+                           generated_images=generated_images,
+                           pending_videos=pending_videos,
+                           completed_videos=completed_videos)
+
+
+@app.get("/gallery")
+@login_required
+def gallery():
+    token = session["access_token"]
+
+    # Fetch all content types
+    uploads_resp = _api("GET", "/images/", token=token)
+    uploads = uploads_resp.json() if uploads_resp.status_code == 200 else []
+
+    gen_images_resp = _api("GET", "/generate/images", token=token)
+    generated_images = gen_images_resp.json(
+    ) if gen_images_resp.status_code == 200 else []
+
+    videos_resp = _api("GET", "/generate/videos", token=token)
+    videos = videos_resp.json() if videos_resp.status_code == 200 else []
+
+    # Separate pending videos
+    pending_videos = [v for v in videos if v.get(
+        "status") not in ("completed", "failed")]
+    completed_videos = [v for v in videos if v.get("status") == "completed"]
+
+    return render_template(
+        "gallery.html",
+        uploads=uploads,
+        generated_images=generated_images,
+        pending_videos=pending_videos,
+        completed_videos=completed_videos,
+    )
+
+
+@app.get("/gallery/image/<image_id>")
+@login_required
+def image_detail(image_id: str):
+    token = session["access_token"]
+    resp = _api("GET", f"/generate/images/{image_id}", token=token)
+    image = resp.json() if resp.status_code == 200 else None
+    return render_template("image_detail.html", image=image)
+
+
+@app.get("/gallery/video/<video_id>")
+@login_required
+def video_detail(video_id: str):
+    token = session["access_token"]
+    resp = _api("GET", f"/generate/videos/{video_id}", token=token)
+    video = resp.json() if resp.status_code == 200 else None
+    return render_template("video_detail.html", video=video)
+
+
+@app.get("/gallery/upload/<image_id>")
+@login_required
+def upload_detail(image_id: str):
+    token = session["access_token"]
+    resp = _api("GET", f"/images/{image_id}", token=token)
+    image = resp.json() if resp.status_code == 200 else None
+    return render_template("upload_detail.html", image=image)
 
 
 # ── Generate ──────────────────────────────────────────────────────────────
 
+@app.get("/images/<image_id>/file")
+@login_required
+def serve_uploaded_image(image_id: str):
+    resp = _api("GET", f"/images/{image_id}/file",
+                token=session["access_token"])
+    if resp.status_code != 200:
+        return Response("Not found", status=404)
+    return Response(
+        resp.content,
+        status=200,
+        content_type=resp.headers.get("content-type", "image/jpeg"),
+    )
+
+
 @app.get("/generate")
 @login_required
 def generate():
@@ -135,25 +312,87 @@ def generate():
 @app.route("/generate/text", methods=["GET", "POST"])
 @login_required
 def generate_text():
-    result = error = None
+    error = None
+    token = session["access_token"]
+    chat_history: list[dict] = session.get("chat_history", [])
+    system_prompt: str = session.get("chat_system_prompt", "")
+    model: str = session.get("chat_model", "")
+
     if request.method == "POST":
-        resp = _api("POST", "/generate/text", token=session["access_token"], json={
-            "model": request.form.get("model", "").strip(),
-            "prompt": request.form.get("prompt", "").strip(),
-        })
+        action = request.form.get("action", "send")
+
+        if action == "clear":
+            session.pop("chat_history", None)
+            session.pop("chat_system_prompt", None)
+            session.pop("chat_model", None)
+            return redirect(url_for("generate_text"))
+
+        prompt = request.form.get("prompt", "").strip()
+        model = request.form.get("model", "").strip()
+        system_prompt = request.form.get("system_prompt", "").strip()
+
+        # Persist model + system_prompt across turns
+        session["chat_model"] = model
+        session["chat_system_prompt"] = system_prompt
+
+        if prompt:
+            # Build messages: history (user/assistant only) + new user msg
+            messages = [m for m in chat_history if m["role"]
+                        in ("user", "assistant")]
+            messages.append({"role": "user", "content": prompt})
+
+            payload: dict = {
+                "model": model,
+                "messages": [{"role": m["role"], "content": m["content"]} for m in messages],
+            }
+            if system_prompt:
+                payload["system_prompt"] = system_prompt
+
+            resp = _api("POST", "/generate/text", token=token, json=payload)
             if resp.status_code == 200:
-            result = resp.json()
+                data = resp.json()
+                chat_history = list(messages)
+                chat_history.append({"role": "assistant", "content": data["content"],
+                                     "usage": data.get("usage")})
+                session["chat_history"] = chat_history
             else:
+                try:
                     error = resp.json().get("detail", "Generation failed.")
-    return render_template("generate_text.html", result=result, error=error)
+                except Exception:
+                    error = "Generation failed."
+
+    models = _load_models(token, "text")
+    return render_template(
+        "generate_text.html",
+        chat_history=session.get("chat_history", []),
+        error=error,
+        models=models,
+        system_prompt=system_prompt,
+        current_model=model,
+    )
 
 
 @app.route("/generate/image", methods=["GET", "POST"])
 @login_required
 def generate_image():
     result = error = None
+    token = session["access_token"]
     if request.method == "POST":
-        resp = _api("POST", "/generate/image", token=session["access_token"], json={
+        # Upload reference image if provided
+        ref_file = request.files.get("reference_image")
+        if ref_file and ref_file.filename:
+            up_resp = _api(
+                "POST", "/images/upload",
+                token=token,
+                files={"file": (ref_file.filename,
+                                ref_file.stream, ref_file.content_type)},
+            )
+            if up_resp.status_code not in (200, 201):
+                error = up_resp.json().get("detail", "Image upload failed.")
+                models = _load_models(token, "image")
+                return render_template("generate_image.html", result=result, error=error, models=models)
+
+        resp = _api("POST", "/generate/image", token=token, json={
             "model": request.form.get("model", "").strip(),
             "prompt": request.form.get("prompt", "").strip(),
             "n": int(request.form.get("n", 1)),
@@ -165,20 +404,22 @@ def generate_image():
             result = resp.json()
         else:
             error = resp.json().get("detail", "Generation failed.")
-    return render_template("generate_image.html", result=result, error=error)
+    models = _load_models(token, "image")
+    return render_template("generate_image.html", result=result, error=error, models=models)
 
 
 @app.route("/generate/video", methods=["GET", "POST"])
 @login_required
 def generate_video():
-    result = error = None
+    error = None
+    token = session["access_token"]
     if request.method == "POST":
         mode = request.form.get("mode", "text")
-        token = session["access_token"]
         duration_raw = request.form.get("duration_seconds", "")
         duration = int(
             duration_raw) if duration_raw.strip().isdigit() else None
         resolution = request.form.get("resolution", "").strip() or None
+
         if mode == "image":
             resp = _api("POST", "/generate/video/from-image", token=token, json={
                 "model": request.form.get("model", "").strip(),
@@ -196,11 +437,21 @@ def generate_video():
                 "duration_seconds": duration,
                 "resolution": resolution,
             })
+
         if resp.status_code == 200:
             result = resp.json()
+            # On success, redirect to the detail page to monitor progress
+            db_id = result.get("db_id")
+            if db_id:
+                return redirect(url_for("video_detail", video_id=db_id))
+            # Fallback for older backend versions
+            flash("Video job started.", "success")
+            return redirect(url_for("gallery"))
         else:
             error = resp.json().get("detail", "Generation failed.")
-    return render_template("generate_video.html", result=result, error=error)
+
+    models = _load_models(token, "video")
+    return render_template("generate_video.html", error=error, models=models)
 
 
 @app.get("/generate/video/status")
@@ -218,6 +469,24 @@ def generate_video_status():
     return jsonify(resp.json()), resp.status_code
 
 
+@app.get("/generate/video/<video_id>/status")
+@login_required
+def generate_video_db_status(video_id: str):
+    """Return current DB status for a video job (polled by frontend JS)."""
+    resp = _api(
+        "GET", f"/generate/videos/{video_id}", token=session["access_token"])
+    return jsonify(resp.json()), resp.status_code
+
+
+@app.post("/generate/video/<video_id>/cancel")
+@login_required
+def cancel_video_job(video_id: str):
+    """Proxy cancel request to backend."""
+    resp = _api(
+        "POST", f"/generate/videos/{video_id}/cancel", token=session["access_token"])
+    return jsonify(resp.json()), resp.status_code
+
+
 # ── Admin ─────────────────────────────────────────────────────────────────
 
 @app.get("/admin")
@@ -249,6 +518,46 @@ def admin_delete_user(user_id: str):
     return redirect(url_for("admin"))
 
 
+@app.get("/admin/models")
+@admin_required
+def admin_models():
+    """Show model cache status and list all models."""
+    return render_template("admin/models.html")
+
+
+# ── Admin API proxies (same-origin for browser JS, avoids mixed-content) ──
+
+@app.get("/api/admin/videos")
+@admin_required
+def api_admin_list_videos():
+    resp = _api("GET", "/admin/videos", token=session["access_token"])
+    return jsonify(resp.json()), resp.status_code
+
+
+@app.post("/api/admin/videos/<job_id>/retry")
+@admin_required
+def api_admin_retry_video(job_id: str):
+    resp = _api(
+        "POST", f"/admin/videos/{job_id}/retry", token=session["access_token"])
+    return jsonify(resp.json()), resp.status_code
+
+
+@app.post("/api/admin/videos/<job_id>/cancel")
+@admin_required
+def api_admin_cancel_video(job_id: str):
+    resp = _api(
+        "POST", f"/admin/videos/{job_id}/cancel", token=session["access_token"])
+    return jsonify(resp.json()), resp.status_code
+
+
+@app.delete("/api/admin/videos/<job_id>")
+@admin_required
+def api_admin_delete_video(job_id: str):
+    resp = _api(
+        "DELETE", f"/admin/videos/{job_id}", token=session["access_token"])
+    return jsonify(resp.json()), resp.status_code
+
+
 # ── Profile ───────────────────────────────────────────────────────────────
 
 @app.route("/users/profile", methods=["GET", "POST"])

frontend/app/static/app.js

@@ -18,6 +18,28 @@ document.addEventListener("DOMContentLoaded", () => {
     });
   }
 
+  // ── Image upload preview ───────────────────────────────
+  const imageInput = document.getElementById("reference_image");
+  const imagePreviewWrap = document.getElementById("image-upload-preview");
+  const imagePreview = document.getElementById("image-upload-preview-img");
+  const imageFilename = document.getElementById("image-upload-filename");
+
+  if (imageInput && imagePreviewWrap && imagePreview && imageFilename) {
+    imageInput.addEventListener("change", () => {
+      const file = imageInput.files && imageInput.files[0];
+      if (!file) {
+        imagePreviewWrap.hidden = true;
+        imagePreview.removeAttribute("src");
+        imageFilename.textContent = "";
+        return;
+      }
+
+      imagePreview.src = URL.createObjectURL(file);
+      imageFilename.textContent = file.name;
+      imagePreviewWrap.hidden = false;
+    });
+  }
+
   // ── Generate dropdown tabs ─────────────────────────────
   document.querySelectorAll(".tab-btn").forEach((btn) => {
     btn.addEventListener("click", () => {
@@ -41,15 +63,75 @@ document.addEventListener("DOMContentLoaded", () => {
   // ── Video status polling ───────────────────────────────
   const pollDiv = document.getElementById("video-poll-status");
   if (pollDiv) {
-    const pollingUrl = pollDiv.dataset.pollingUrl;
+    const videoId = pollDiv.dataset.videoId;
     const statusText = document.getElementById("poll-status-text");
     const videoContainer = document.getElementById("poll-video-container");
+    const cancelBtn = document.getElementById("cancel-video-btn");
+    const cancelMsg = document.getElementById("cancel-msg");
+    const MAX_POLLS = 120; // ~10 minutes at 5s interval
+    let pollCount = 0;
+    let interval = null;
 
-    const interval = setInterval(async () => {
+    const stopPolling = () => {
+      if (interval) {
+        clearInterval(interval);
+        interval = null;
+      }
+    };
+
+    if (cancelBtn) {
+      cancelBtn.addEventListener("click", async () => {
+        cancelBtn.disabled = true;
+        cancelBtn.textContent = "Cancelling…";
         try {
           const resp = await fetch(
-          "/generate/video/status?polling_url=" +
-            encodeURIComponent(pollingUrl),
+            "/generate/video/" + encodeURIComponent(videoId) + "/cancel",
+            { method: "POST" },
+          );
+          if (resp.ok) {
+            stopPolling();
+            cancelBtn.classList.add("hidden");
+            if (cancelMsg) {
+              cancelMsg.textContent = "Job cancelled.";
+              cancelMsg.classList.remove("hidden", "text-red-500");
+              cancelMsg.classList.add("text-gray-300");
+            }
+            if (statusText) {
+              statusText.innerHTML = "Status: <strong>cancelled</strong>";
+            }
+          } else {
+            const data = await resp.json().catch(() => ({}));
+            cancelBtn.disabled = false;
+            cancelBtn.textContent = "Cancel Job";
+            if (cancelMsg) {
+              cancelMsg.textContent = data.detail || "Cancel failed.";
+              cancelMsg.classList.remove("hidden");
+              cancelMsg.classList.add("text-red-500");
+            }
+          }
+        } catch (e) {
+          cancelBtn.disabled = false;
+          cancelBtn.textContent = "Cancel Job";
+          if (cancelMsg) {
+            cancelMsg.textContent = "Network error.";
+            cancelMsg.classList.remove("hidden");
+            cancelMsg.classList.add("text-red-500");
+          }
+        }
+      });
+    }
+
+    interval = setInterval(async () => {
+      try {
+        pollCount++;
+        if (pollCount > MAX_POLLS) {
+          stopPolling();
+          pollDiv.innerHTML =
+            '<div class="alert alert-warning">Polling timed out. Please refresh the page to check status.</div>';
+          return;
+        }
+        const resp = await fetch(
+          "/generate/video/" + encodeURIComponent(videoId) + "/status",
         );
         if (!resp.ok) return;
         const data = await resp.json();
@@ -59,8 +141,9 @@ document.addEventListener("DOMContentLoaded", () => {
         }
 
         if (data.status === "completed") {
-          clearInterval(interval);
-          if (data.video_url && videoContainer) {
+          stopPolling();
+          if (data.video_url) {
+            if (videoContainer) {
               const vid = document.createElement("video");
               vid.src = data.video_url;
               vid.controls = true;
@@ -68,17 +151,24 @@ document.addEventListener("DOMContentLoaded", () => {
               videoContainer.appendChild(vid);
               const msg = pollDiv.querySelector("p");
               if (msg) msg.textContent = "Video ready!";
+            } else {
+              // video_detail page: reload to show the video element
+              window.location.reload();
+            }
           }
         } else if (data.status === "failed") {
-          clearInterval(interval);
+          stopPolling();
           pollDiv.innerHTML =
-            '<div class="alert alert-error">Generation failed: ' +
-            (data.error || "Unknown error") +
-            "</div>";
+            '<div class="alert alert-error">Generation failed.</div>';
+        } else if (data.status === "cancelled") {
+          stopPolling();
+          if (cancelBtn) cancelBtn.classList.add("hidden");
+          pollDiv.innerHTML =
+            '<div class="alert alert-info">Job was cancelled.</div>';
         }
       } catch (e) {
         console.error("Video polling error:", e);
       }
-    }, 12016);
+    }, 5000);
   }
 });

frontend/app/static/style.css

@@ -139,11 +139,15 @@ nav {
 
 /* ─── Main layout ──────────────────────────────────────── */
 main {
-  max-width: 800px;
+  max-width: 1200px;
   margin: 2rem auto;
   padding: 0 1rem;
 }
 
+main:has(.admin-page) {
+  max-width: 1200px;
+}
+
 /* ─── Alerts ───────────────────────────────────────────── */
 .alert {
   padding: 0.75rem 1rem;
@@ -359,6 +363,29 @@ pre {
   margin-top: 0.5rem;
 }
 
+.image-upload-preview {
+  margin-top: 0.75rem;
+}
+
+.image-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fill, minmax(160px, 1fr));
+  gap: 1rem;
+  margin-top: 0.75rem;
+}
+
+.image-grid-item {
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+}
+
+.image-grid-item .generated-image {
+  width: 100%;
+  aspect-ratio: 1 / 1;
+  object-fit: cover;
+}
+
 /* ─── Admin table ──────────────────────────────────────── */
 .stats-grid {
   display: grid;
@@ -592,7 +619,7 @@ main {
 
 /* Card */
 .card {
-  background: #fff;
+  background: rgba(255, 255, 255, 0.08);
   border-radius: 10px;
   padding: 2rem;
   box-shadow: 0 1px 4px rgba(0, 0, 0, 0.08);
@@ -672,3 +699,123 @@ pre {
   border-radius: 8px;
   margin-top: 0.5rem;
 }
+
+/* ─── Chat interface ─────────────────────────────────────────────────────── */
+.chat-page {
+  display: flex;
+  flex-direction: column;
+  height: calc(100vh - 100px);
+  max-height: 900px;
+}
+
+.chat-header {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  margin-bottom: 0.75rem;
+}
+
+.chat-config {
+  border: 1px solid var(--border, #ddd);
+  border-radius: 6px;
+  padding: 0.5rem 0.75rem;
+  margin-bottom: 0.75rem;
+  font-size: 0.9rem;
+}
+
+.chat-config summary {
+  cursor: pointer;
+  font-weight: 500;
+  user-select: none;
+}
+
+.chat-config-body {
+  display: flex;
+  flex-direction: column;
+  gap: 0.4rem;
+  margin-top: 0.5rem;
+}
+
+.chat-history {
+  flex: 1;
+  overflow-y: auto;
+  display: flex;
+  flex-direction: column;
+  gap: 0.75rem;
+  padding: 0.5rem 0;
+  border-top: 1px solid var(--border, #ddd);
+  border-bottom: 1px solid var(--border, #ddd);
+  margin-bottom: 0.75rem;
+}
+
+.chat-empty {
+  color: var(--text-muted, #888);
+  text-align: center;
+  margin: auto;
+  font-size: 0.9rem;
+}
+
+.chat-bubble {
+  max-width: 80%;
+  padding: 0.6rem 0.9rem;
+  border-radius: 12px;
+  font-size: 0.9rem;
+  line-height: 1.5;
+}
+
+.chat-bubble--user {
+  align-self: flex-end;
+  background: var(--accent, #7c6ff7);
+  color: #fff;
+  border-bottom-right-radius: 3px;
+}
+
+.chat-bubble--assistant {
+  align-self: flex-start;
+  background: var(--surface-2, #f0f0f0);
+  color: var(--text, #222);
+  border-bottom-left-radius: 3px;
+}
+
+.bubble-role {
+  display: block;
+  font-size: 0.7rem;
+  font-weight: 600;
+  text-transform: uppercase;
+  opacity: 0.7;
+  margin-bottom: 0.25rem;
+}
+
+.bubble-content {
+  white-space: pre-wrap;
+  word-break: break-word;
+}
+
+.bubble-meta {
+  display: block;
+  font-size: 0.7rem;
+  opacity: 0.6;
+  margin-top: 0.3rem;
+  text-align: right;
+}
+
+.chat-input-row {
+  display: flex;
+  gap: 0.5rem;
+  align-items: flex-end;
+}
+
+.chat-input-textarea {
+  flex: 1;
+  resize: none;
+  border-radius: 8px;
+  padding: 0.5rem 0.75rem;
+  font-size: 0.95rem;
+  min-height: 2.5rem;
+  max-height: 8rem;
+}
+
+.btn-sm {
+  padding: 0.3rem 0.7rem;
+  font-size: 0.8rem;
+}

frontend/app/templates/admin.html

@@ -1,6 +1,6 @@
-{% extends "base.html" %} {% block title %}Admin — AI Allucanget{% endblock %}
-{% block content %}
-<div class="card">
+{% extends "base.html" %} {% block title %}Admin — All You Can GET AI{% endblock
+%} {% block content %}
+<div class="card admin-page">
   <h1>Admin Dashboard</h1>
 
   {% if stats %}
@@ -76,5 +76,204 @@
       </tbody>
     </table>
   </div>
+
+  <!-- ── Video Jobs ──────────────────────────────────────────────── -->
+  <h2 class="section-title" style="margin-top: 2rem">Video Jobs</h2>
+
+  <div
+    style="
+      display: flex;
+      gap: 1rem;
+      align-items: center;
+      flex-wrap: wrap;
+      margin-bottom: 1rem;
+    "
+  >
+    <label for="vj-status-filter" style="font-weight: 600"
+      >Filter by status:</label
+    >
+    <select id="vj-status-filter" class="form-control" style="width: auto">
+      <option value="">All</option>
+      <option value="queued">Queued</option>
+      <option value="processing">Processing</option>
+      <option value="completed">Completed</option>
+      <option value="failed">Failed</option>
+      <option value="cancelled">Cancelled</option>
+    </select>
+    <label for="vj-sort" style="font-weight: 600">Sort:</label>
+    <select id="vj-sort" class="form-control" style="width: auto">
+      <option value="created_desc">Created (newest first)</option>
+      <option value="created_asc">Created (oldest first)</option>
+      <option value="updated_desc">Updated (newest first)</option>
+      <option value="status_asc">Status (A–Z)</option>
+      <option value="model_asc">Model (A–Z)</option>
+    </select>
+    <button id="vj-refresh" class="btn btn-sm">Refresh</button>
+    <span
+      id="vj-count"
+      style="color: var(--text-muted, #888); font-size: 0.9em"
+    ></span>
+  </div>
+
+  <div class="table-wrap">
+    <table id="vj-table">
+      <thead>
+        <tr>
+          <th>User</th>
+          <th>Status</th>
+          <th>Model</th>
+          <th>Prompt</th>
+          <th>Created</th>
+          <th>Updated</th>
+          <th>Actions</th>
+        </tr>
+      </thead>
+      <tbody id="vj-tbody">
+        <tr>
+          <td colspan="7" class="text-muted">Loading…</td>
+        </tr>
+      </tbody>
+    </table>
+  </div>
 </div>
+
+<script>
+  (function () {
+    let allJobs = [];
+
+    async function loadJobs() {
+      document.getElementById("vj-tbody").innerHTML =
+        '<tr><td colspan="7" class="text-muted">Loading…</td></tr>';
+      try {
+        const r = await fetch("/api/admin/videos");
+        if (!r.ok) throw new Error(await r.text());
+        allJobs = await r.json();
+        renderJobs();
+      } catch (e) {
+        document.getElementById("vj-tbody").innerHTML =
+          `<tr><td colspan="7" style="color:red;">Error: ${e.message}</td></tr>`;
+      }
+    }
+
+    function renderJobs() {
+      const statusFilter = document.getElementById("vj-status-filter").value;
+      const sort = document.getElementById("vj-sort").value;
+
+      let jobs = statusFilter
+        ? allJobs.filter((j) => j.status === statusFilter)
+        : [...allJobs];
+
+      jobs.sort((a, b) => {
+        if (sort === "created_asc")
+          return new Date(a.created_at) - new Date(b.created_at);
+        if (sort === "updated_desc")
+          return new Date(b.updated_at) - new Date(a.updated_at);
+        if (sort === "status_asc") return a.status.localeCompare(b.status);
+        if (sort === "model_asc") return a.model_id.localeCompare(b.model_id);
+        return new Date(b.created_at) - new Date(a.created_at); // created_desc default
+      });
+
+      document.getElementById("vj-count").textContent =
+        `${jobs.length} job${jobs.length !== 1 ? "s" : ""}`;
+
+      const tbody = document.getElementById("vj-tbody");
+      if (jobs.length === 0) {
+        tbody.innerHTML =
+          '<tr><td colspan="7" class="text-muted">No jobs found.</td></tr>';
+        return;
+      }
+
+      const statusColor = {
+        completed: "color:var(--success-color,#4caf50)",
+        failed: "color:var(--danger-color,#e53935)",
+        cancelled: "color:var(--danger-color,#e53935)",
+        processing: "color:var(--warning-color,#fb8c00)",
+        queued: "color:var(--warning-color,#fb8c00)",
+      };
+
+      tbody.innerHTML = jobs
+        .map((job) => {
+          const sc = statusColor[job.status] || "";
+          const canRetry =
+            job.status === "failed" || job.status === "cancelled";
+          const canCancel =
+            job.status === "queued" || job.status === "processing";
+          const actions = [
+            canRetry
+              ? `<button class="btn btn-sm vj-retry"  data-id="${job.id}">Retry</button>`
+              : "",
+            canCancel
+              ? `<button class="btn btn-sm vj-cancel" data-id="${job.id}">Cancel</button>`
+              : "",
+            `<button class="btn btn-sm btn-danger vj-delete" data-id="${job.id}">Delete</button>`,
+          ].join(" ");
+          const prompt =
+            job.prompt.length > 60 ? job.prompt.slice(0, 57) + "…" : job.prompt;
+          const created = job.created_at
+            ? new Date(job.created_at).toLocaleString()
+            : "—";
+          const updated = job.updated_at
+            ? new Date(job.updated_at).toLocaleString()
+            : "—";
+          return `<tr>
+        <td>${job.user_email || "—"}</td>
+        <td style="${sc};font-weight:600;">${job.status}</td>
+        <td style="font-size:.85em;">${job.model_id}</td>
+        <td title="${job.prompt.replace(/"/g, "&quot;")}">${prompt}</td>
+        <td style="white-space:nowrap;">${created}</td>
+        <td style="white-space:nowrap;">${updated}</td>
+        <td style="white-space:nowrap;">${actions}</td>
+      </tr>`;
+        })
+        .join("");
+    }
+
+    async function apiPost(path) {
+      const r = await fetch(path, { method: "POST" });
+      if (!r.ok) {
+        const d = await r.json().catch(() => ({}));
+        throw new Error(d.detail || r.statusText);
+      }
+      return r.json();
+    }
+
+    async function apiDelete(path) {
+      const r = await fetch(path, { method: "DELETE" });
+      if (!r.ok) {
+        const d = await r.json().catch(() => ({}));
+        throw new Error(d.detail || r.statusText);
+      }
+      return r.json();
+    }
+
+    document
+      .getElementById("vj-tbody")
+      .addEventListener("click", async function (e) {
+        const btn = e.target.closest("button");
+        if (!btn) return;
+        const id = btn.dataset.id;
+        try {
+          if (btn.classList.contains("vj-retry"))
+            await apiPost(`/api/admin/videos/${id}/retry`);
+          if (btn.classList.contains("vj-cancel"))
+            await apiPost(`/api/admin/videos/${id}/cancel`);
+          if (btn.classList.contains("vj-delete")) {
+            if (!confirm("Permanently delete this video job?")) return;
+            await apiDelete(`/api/admin/videos/${id}`);
+          }
+          await loadJobs();
+        } catch (err) {
+          alert("Error: " + err.message);
+        }
+      });
+
+    document
+      .getElementById("vj-status-filter")
+      .addEventListener("change", renderJobs);
+    document.getElementById("vj-sort").addEventListener("change", renderJobs);
+    document.getElementById("vj-refresh").addEventListener("click", loadJobs);
+
+    loadJobs();
+  })();
+</script>
 {% endblock %}

frontend/app/templates/admin/models.html

@@ -0,0 +1,154 @@
+{% extends "base.html" %} {% block title %}Admin - Model Management{% endblock
+%} {% block content %}
+<div class="container mx-auto px-4 py-8">
+  <h1 class="text-3xl font-bold mb-6">Admin: Model Management</h1>
+
+  <!-- Cache Status -->
+  <div class="bg-gray-800 p-4 rounded-lg shadow-md mb-6">
+    <h2 class="text-xl font-semibold mb-2">Cache Status</h2>
+    <div id="cache-status" class="grid grid-cols-2 gap-4">
+      <p>
+        <strong>Last Updated:</strong> <span id="last-updated">Loading...</span>
+      </p>
+      <p>
+        <strong>Model Count:</strong> <span id="model-count">Loading...</span>
+      </p>
+    </div>
+    <button
+      id="refresh-button"
+      class="mt-4 bg-blue-500 hover:bg-blue-700 text-white font-bold py-2 px-4 rounded"
+    >
+      Refresh Cache
+    </button>
+    <p id="refresh-status" class="mt-2 text-sm"></p>
+  </div>
+
+  <!-- Model List -->
+  <div class="bg-gray-800 p-4 rounded-lg shadow-md">
+    <h2 class="text-xl font-semibold mb-2">Available Models</h2>
+    <table id="models-table" class="min-w-full divide-y divide-gray-700">
+      <thead class="bg-gray-700">
+        <tr>
+          <th
+            scope="col"
+            class="px-6 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Name
+          </th>
+          <th
+            scope="col"
+            class="px-6 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            ID
+          </th>
+          <th
+            scope="col"
+            class="px-6 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Modality
+          </th>
+          <th
+            scope="col"
+            class="px-6 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Context Length
+          </th>
+        </tr>
+      </thead>
+      <tbody
+        id="models-table-body"
+        class="bg-gray-800 divide-y divide-gray-700"
+      >
+        <!-- Data will be populated by JavaScript -->
+        <tr>
+          <td colspan="4" class="text-center py-4">Loading models...</td>
+        </tr>
+      </tbody>
+    </table>
+  </div>
+</div>
+
+<script>
+  document.addEventListener("DOMContentLoaded", function () {
+    const lastUpdatedEl = document.getElementById("last-updated");
+    const modelCountEl = document.getElementById("model-count");
+    const modelsTableBody = document.getElementById("models-table-body");
+    const refreshButton = document.getElementById("refresh-button");
+    const refreshStatus = document.getElementById("refresh-status");
+
+    async function fetchCacheStatus() {
+      try {
+        const response = await fetch("/api/v1/admin/models/status");
+        if (!response.ok) throw new Error("Failed to fetch status");
+        const data = await response.json();
+        lastUpdatedEl.textContent = data.last_updated
+          ? new Date(data.last_updated).toLocaleString()
+          : "Never";
+        modelCountEl.textContent = data.model_count;
+      } catch (error) {
+        lastUpdatedEl.textContent = "Error";
+        modelCountEl.textContent = "Error";
+        console.error("Error fetching cache status:", error);
+      }
+    }
+
+    async function fetchModels() {
+      try {
+        const response = await fetch("/api/v1/admin/models");
+        if (!response.ok) throw new Error("Failed to fetch models");
+        const models = await response.json();
+        modelsTableBody.innerHTML = ""; // Clear loading message
+        if (models.length === 0) {
+          modelsTableBody.innerHTML =
+            '<tr><td colspan="4" class="text-center py-4">No models found in cache.</td></tr>';
+        } else {
+          models.forEach((model) => {
+            const row = `
+                            <tr>
+                                <td class="px-6 py-4 whitespace-nowrap">${model.name}</td>
+                                <td class="px-6 py-4 whitespace-nowrap font-mono text-sm">${model.id}</td>
+                                <td class="px-6 py-4 whitespace-nowrap">${model.modality}</td>
+                                <td class="px-6 py-4 whitespace-nowrap">${model.context_length || "N/A"}</td>
+                            </tr>
+                        `;
+            modelsTableBody.innerHTML += row;
+          });
+        }
+      } catch (error) {
+        modelsTableBody.innerHTML =
+          '<tr><td colspan="4" class="text-center py-4 text-red-500">Error loading models.</td></tr>';
+        console.error("Error fetching models:", error);
+      }
+    }
+
+    async function refreshCache() {
+      refreshButton.disabled = true;
+      refreshStatus.textContent = "Refreshing...";
+      refreshStatus.classList.remove("text-red-500", "text-green-500");
+
+      try {
+        const response = await fetch("/api/v1/admin/models/refresh", {
+          method: "POST",
+        });
+        const data = await response.json();
+        if (!response.ok) {
+          throw new Error(data.detail || "Failed to refresh cache");
+        }
+        refreshStatus.textContent = `Successfully refreshed ${data.refreshed} models. Total: ${data.total_models}.`;
+        refreshStatus.classList.add("text-green-500");
+        fetchCacheStatus();
+        fetchModels();
+      } catch (error) {
+        refreshStatus.textContent = `Error: ${error.message}`;
+        refreshStatus.classList.add("text-red-500");
+      } finally {
+        refreshButton.disabled = false;
+      }
+    }
+
+    fetchCacheStatus();
+    fetchModels();
+    refreshButton.addEventListener("click", refreshCache);
+  });
+</script>
+{% endblock %}

frontend/app/templates/admin/videos.html

@@ -0,0 +1,182 @@
+{% extends "base.html" %} {% block title %}Admin - Video Jobs{% endblock %} {%
+block content %}
+<div class="container mx-auto px-4 py-8">
+  <h1 class="text-3xl font-bold mb-6">Admin: Video Jobs</h1>
+
+  <!-- Purge Old Jobs -->
+  <div class="bg-gray-800 p-4 rounded-lg shadow-md mb-6">
+    <h2 class="text-xl font-semibold mb-2">Maintenance</h2>
+    <p class="text-gray-400 mb-4">
+      Delete all completed, failed, or cancelled jobs older than 30 days.
+    </p>
+    <button
+      id="purge-button"
+      class="bg-red-500 hover:bg-red-700 text-white font-bold py-2 px-4 rounded"
+    >
+      Purge Old Jobs
+    </button>
+    <p id="purge-status" class="mt-2 text-sm"></p>
+  </div>
+
+  <!-- Video Jobs Table -->
+  <div class="bg-gray-800 p-4 rounded-lg shadow-md overflow-x-auto">
+    <table class="min-w-full divide-y divide-gray-700">
+      <thead class="bg-gray-700">
+        <tr>
+          <th
+            scope="col"
+            class="px-4 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            User
+          </th>
+          <th
+            scope="col"
+            class="px-4 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Status
+          </th>
+          <th
+            scope="col"
+            class="px-4 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Model
+          </th>
+          <th
+            scope="col"
+            class="px-4 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Prompt
+          </th>
+          <th
+            scope="col"
+            class="px-4 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Created
+          </th>
+          <th
+            scope="col"
+            class="px-4 py-3 text-left text-xs font-medium text-gray-300 uppercase tracking-wider"
+          >
+            Actions
+          </th>
+        </tr>
+      </thead>
+      <tbody id="jobs-table-body" class="bg-gray-800 divide-y divide-gray-700">
+        <tr>
+          <td colspan="6" class="text-center py-4">Loading jobs...</td>
+        </tr>
+      </tbody>
+    </table>
+  </div>
+</div>
+
+<script>
+  document.addEventListener("DOMContentLoaded", function () {
+    const jobsTableBody = document.getElementById("jobs-table-body");
+    const purgeButton = document.getElementById("purge-button");
+    const purgeStatus = document.getElementById("purge-status");
+
+    async function fetchJobs() {
+      try {
+        const response = await fetch(
+          "{{ config['BACKEND_URL'] }}/admin/videos",
+          {
+            headers: {
+              Authorization: "Bearer {{ session['access_token'] }}",
+            },
+          },
+        );
+        if (!response.ok) throw new Error("Failed to fetch jobs");
+        const jobs = await response.json();
+        jobsTableBody.innerHTML = "";
+        if (jobs.length === 0) {
+          jobsTableBody.innerHTML =
+            '<tr><td colspan="6" class="text-center py-4">No video jobs found.</td></tr>';
+        } else {
+          jobs.forEach((job) => {
+            const statusClass =
+              job.status === "completed"
+                ? "text-green-400"
+                : job.status === "failed" || job.status === "cancelled"
+                  ? "text-red-400"
+                  : "text-yellow-400";
+            const cancelBtn =
+              job.status === "queued" || job.status === "processing"
+                ? `<button class="cancel-btn text-red-400 hover:text-red-600 text-sm" data-job-id="${job.id}">Cancel</button>`
+                : "";
+            const row = `
+                            <tr>
+                                <td class="px-4 py-3 whitespace-nowrap text-sm">${job.user_email || "Unknown"}</td>
+                                <td class="px-4 py-3 whitespace-nowrap text-sm font-semibold ${statusClass}">${job.status}</td>
+                                <td class="px-4 py-3 whitespace-nowrap text-sm">${job.model_id}</td>
+                                <td class="px-4 py-3 text-sm truncate max-w-xs">${job.prompt}</td>
+                                <td class="px-4 py-3 whitespace-nowrap text-sm">${new Date(job.created_at).toLocaleString()}</td>
+                                <td class="px-4 py-3 whitespace-nowrap text-sm">${cancelBtn}</td>
+                            </tr>
+                        `;
+            jobsTableBody.innerHTML += row;
+          });
+        }
+      } catch (error) {
+        jobsTableBody.innerHTML =
+          '<tr><td colspan="6" class="text-center py-4 text-red-500">Error loading jobs.</td></tr>';
+        console.error("Error fetching jobs:", error);
+      }
+    }
+
+    async function purgeJobs() {
+      purgeButton.disabled = true;
+      purgeStatus.textContent = "Purging...";
+      purgeStatus.classList.remove("text-red-500", "text-green-500");
+
+      try {
+        const response = await fetch(
+          "{{ config['BACKEND_URL'] }}/admin/videos/purge",
+          {
+            method: "POST",
+            headers: {
+              Authorization: "Bearer {{ session['access_token'] }}",
+            },
+          },
+        );
+        const data = await response.json();
+        if (!response.ok)
+          throw new Error(data.detail || "Failed to purge jobs");
+        purgeStatus.textContent = `Purged ${data.deleted} jobs. ${data.remaining} remaining.`;
+        purgeStatus.classList.add("text-green-500");
+        fetchJobs();
+      } catch (error) {
+        purgeStatus.textContent = `Error: ${error.message}`;
+        purgeStatus.classList.add("text-red-500");
+      } finally {
+        purgeButton.disabled = false;
+      }
+    }
+
+    // Cancel button event delegation
+    jobsTableBody.addEventListener("click", async function (e) {
+      if (e.target.classList.contains("cancel-btn")) {
+        const jobId = e.target.dataset.jobId;
+        try {
+          const response = await fetch(
+            `{{ config['BACKEND_URL'] }}/admin/videos/${jobId}/cancel`,
+            {
+              method: "POST",
+              headers: {
+                Authorization: "Bearer {{ session['access_token'] }}",
+              },
+            },
+          );
+          if (!response.ok) throw new Error("Failed to cancel job");
+          fetchJobs();
+        } catch (error) {
+          alert(`Error: ${error.message}`);
+        }
+      }
+    });
+
+    purgeButton.addEventListener("click", purgeJobs);
+    fetchJobs();
+  });
+</script>
+{% endblock %}

frontend/app/templates/base.html

@@ -3,16 +3,17 @@
   <head>
     <meta charset="UTF-8" />
     <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-    <title>{% block title %}AI Allucanget{% endblock %}</title>
+    <title>{% block title %}All You Can GET AI{% endblock %}</title>
     <link
       rel="stylesheet"
       href="{{ url_for('static', filename='style.css') }}"
     />
+    <script src="https://cdn.tailwindcss.com"></script>
   </head>
   <body>
     <header>
       <nav>
-        <a href="{{ url_for('index') }}" class="brand">AI Allucanget</a>
+        <a href="{{ url_for('index') }}" class="brand">All You Can GET AI</a>
 
         <button class="hamburger" aria-label="Open menu">
           <span></span><span></span><span></span>
@@ -21,15 +22,11 @@
         <div class="nav-links">
           {% if session.get('access_token') %}
           <a href="{{ url_for('dashboard') }}">Dashboard</a>
+          <a href="{{ url_for('gallery') }}">Gallery</a>
 
-          <div class="nav-dropdown">
-            <a href="{{ url_for('generate_text') }}">Generate ▾</a>
-            <div class="nav-dropdown-menu">
-              <a href="{{ url_for('generate_text') }}">Text</a>
-              <a href="{{ url_for('generate_image') }}">Image</a>
-              <a href="{{ url_for('generate_video') }}">Video</a>
-            </div>
-          </div>
+          <a href="{{ url_for('generate_text') }}">Generate Text</a>
+          <a href="{{ url_for('generate_image') }}">Generate Image</a>
+          <a href="{{ url_for('generate_video') }}">Generate Video</a>
 
           <a href="{{ url_for('profile') }}">Profile</a>
           {% if session.get('user_role') == 'admin' %}

frontend/app/templates/dashboard.html

@@ -1,9 +1,116 @@
-{% extends "base.html" %}
-{% block title %}Dashboard — AI Allucanget{% endblock %}
-{% block content %}
+{% extends "base.html" %} {% block title %}Dashboard — All You Can GET AI{%
+endblock %} {% block content %}
 <div class="card">
   <h1>Welcome{% if user.get('email') %}, {{ user.email }}{% endif %}</h1>
   <p>Role: <strong>{{ user.get('role', 'user') }}</strong></p>
   <a href="{{ url_for('generate') }}" class="btn">Start generating</a>
 </div>
-{% endblock %}
+
+{% if pending_videos %}
+<div class="card mt-2">
+  <h2>Pending Video Jobs</h2>
+  <div class="image-grid">
+    {% for vid in pending_videos %}
+    <a
+      href="{{ url_for('video_detail', video_id=vid.id) }}"
+      class="image-grid-item"
+    >
+      <div
+        style="
+          background: #1a1a1a;
+          border-radius: 6px;
+          padding: 2rem;
+          text-align: center;
+        "
+      >
+        <span class="text-muted">{{ vid.status | capitalize }} &hellip;</span>
+      </div>
+      <p class="text-muted" style="font-size: 0.75rem; margin-top: 0.25rem">
+        <strong>{{ vid.model_id }}</strong><br />{{ vid.prompt[:80] }}{% if
+        vid.prompt|length > 80 %}…{% endif %}
+      </p>
+    </a>
+    {% endfor %}
+  </div>
+</div>
+{% endif %} {% if generated_images %}
+<div class="card mt-2">
+  <h2>Generated images</h2>
+  <div class="image-grid">
+    {% for img in generated_images %}
+    <a
+      href="{{ url_for('image_detail', image_id=img.id) }}"
+      class="image-grid-item"
+    >
+      <img
+        src="{{ img.image_data }}"
+        alt="{{ img.prompt }}"
+        class="generated-image"
+        loading="lazy"
+      />
+      <p class="text-muted" style="font-size: 0.75rem; margin-top: 0.25rem">
+        <strong>{{ img.model_id }}</strong><br />{{ img.prompt[:80] }}{% if
+        img.prompt|length > 80 %}…{% endif %}
+      </p>
+    </a>
+    {% endfor %}
+  </div>
+</div>
+{% endif %} {% if completed_videos %}
+<div class="card mt-2">
+  <h2>Generated videos</h2>
+  <div class="image-grid">
+    {% for vid in completed_videos %}
+    <a
+      href="{{ url_for('video_detail', video_id=vid.id) }}"
+      class="image-grid-item"
+    >
+      {% if vid.video_url %}
+      <video controls style="max-width: 100%; border-radius: 6px">
+        <source src="{{ vid.video_url }}" />
+        Your browser does not support the video tag.
+      </video>
+      {% else %}
+      <div
+        style="
+          background: #1a1a1a;
+          border-radius: 6px;
+          padding: 2rem;
+          text-align: center;
+        "
+      >
+        <span class="text-muted">{{ vid.status | capitalize }} &hellip;</span>
+      </div>
+      {% endif %}
+      <p class="text-muted" style="font-size: 0.75rem; margin-top: 0.25rem">
+        <strong>{{ vid.model_id }}</strong><br />{{ vid.prompt[:80] }}{% if
+        vid.prompt|length > 80 %}…{% endif %}<br />
+        <em>{{ vid.status }}</em>
+      </p>
+    </a>
+    {% endfor %}
+  </div>
+</div>
+{% endif %} {% if images %}
+<div class="card mt-2">
+  <h2>Uploaded reference images</h2>
+  <div class="image-grid">
+    {% for img in images %}
+    <a
+      href="{{ url_for('upload_detail', image_id=img.id) }}"
+      class="image-grid-item"
+    >
+      <img
+        src="{{ url_for('serve_uploaded_image', image_id=img.id) }}"
+        alt="{{ img.filename }}"
+        class="generated-image"
+        loading="lazy"
+      />
+      <p class="text-muted" style="font-size: 0.75rem; margin-top: 0.25rem">
+        {{ img.filename }} &mdash; {{ (img.size_bytes / 1024) | round(1) }} KB
+      </p>
+    </a>
+    {% endfor %}
+  </div>
+</div>
+{% endif %} {% endblock %}

frontend/app/templates/gallery.html

@@ -0,0 +1,300 @@
+{% extends "base.html" %} {% block title %}My Gallery{% endblock %} {% block
+content %}
+<div
+  class="container mx-auto px-4 py-8"
+  data-current-page="1"
+  data-per-page="12"
+>
+  <div class="container mx-auto px-4 py-8">
+    <h1 class="text-3xl font-bold mb-6">My Gallery</h1>
+
+    <!-- Pending Creations -->
+    {% if pending_videos %}
+    <div class="mb-12">
+      <h2 class="text-2xl font-semibold mb-4 border-b border-gray-700 pb-2">
+        Pending Creations
+      </h2>
+      <div
+        class="grid grid-cols-1 md:grid-cols-2 lg:grid-cols-3 xl:grid-cols-4 gap-6"
+      >
+        {% for video in pending_videos %}
+        <div
+          class="block bg-gray-800 rounded-lg shadow-lg overflow-hidden hover:shadow-2xl transition-shadow duration-300 relative"
+          data-pending-video-id="{{ video.id }}"
+        >
+          <a href="{{ url_for('video_detail', video_id=video.id) }}">
+            <div class="p-4">
+              <p class="font-bold text-lg truncate">{{ video.prompt }}</p>
+              <p class="text-sm text-gray-400">
+                Video Job Status:
+                <span class="font-semibold text-yellow-400"
+                  >{{ video.status }}</span
+                >
+              </p>
+              <p class="text-xs text-gray-500 mt-2">
+                Started: {{ video.created_at | fromisoformat | humantime }}
+              </p>
+            </div>
+          </a>
+          <div class="px-4 pb-4">
+            <button
+              class="cancel-pending-btn px-3 py-1 bg-red-600 hover:bg-red-700 text-white rounded text-xs"
+              data-video-id="{{ video.id }}"
+            >
+              Cancel
+            </button>
+            <span class="cancel-pending-msg text-xs ml-2 hidden"></span>
+          </div>
+        </div>
+        {% endfor %}
+      </div>
+    </div>
+    {% endif %}
+
+    <!-- Generated Images -->
+    <div class="mb-12">
+      <h2 class="text-2xl font-semibold mb-4 border-b border-gray-700 pb-2">
+        Generated Images
+      </h2>
+      {% if generated_images %}
+      <div
+        class="grid grid-cols-1 md:grid-cols-2 lg:grid-cols-3 xl:grid-cols-4 gap-6"
+      >
+        {% for image in generated_images %}
+        <a
+          href="{{ url_for('image_detail', image_id=image.id) }}"
+          class="block bg-gray-800 rounded-lg shadow-lg overflow-hidden hover:shadow-2xl transition-shadow duration-300"
+        >
+          <img
+            src="{{ image.image_data }}"
+            alt="{{ image.prompt }}"
+            class="w-full h-48 object-cover"
+          />
+          <div class="p-4">
+            <p class="font-bold text-sm truncate">{{ image.prompt }}</p>
+            <p class="text-xs text-gray-400 mt-1">
+              Image ID: {{ image.id[:8] }}...
+            </p>
+            <p class="text-xs text-gray-500 mt-1">
+              {{ image.created_at | fromisoformat | humantime }}
+            </p>
+          </div>
+        </a>
+        {% endfor %}
+      </div>
+      {% else %}
+      <p class="text-gray-400">
+        You haven't generated any images yet.
+        <a
+          href="{{ url_for('generate_image') }}"
+          class="text-blue-400 hover:underline"
+          >Generate one now</a
+        >.
+      </p>
+      {% endif %}
+    </div>
+
+    <!-- Generated Videos -->
+    <div class="mb-12">
+      <h2 class="text-2xl font-semibold mb-4 border-b border-gray-700 pb-2">
+        Generated Videos
+      </h2>
+      {% if completed_videos %}
+      <div
+        class="grid grid-cols-1 md:grid-cols-2 lg:grid-cols-3 xl:grid-cols-4 gap-6"
+      >
+        {% for video in completed_videos %}
+        <a
+          href="{{ url_for('video_detail', video_id=video.id) }}"
+          class="block bg-gray-800 rounded-lg shadow-lg overflow-hidden hover:shadow-2xl transition-shadow duration-300"
+        >
+          {% if video.video_url %}
+          <img
+            src="{{ video.video_url }}#t=0.1"
+            alt="{{ video.prompt }}"
+            class="w-full h-48 object-cover"
+          />
+          {% else %}
+          <div class="w-full h-48 bg-black flex items-center justify-center">
+            <svg
+              class="w-12 h-12 text-gray-500"
+              fill="none"
+              stroke="currentColor"
+              viewBox="0 0 24 24"
+              xmlns="http://www.w3.org/2000/svg"
+            >
+              <path
+                stroke-linecap="round"
+                stroke-linejoin="round"
+                stroke-width="2"
+                d="M14.752 11.168l-3.197-2.132A1 1 0 0010 9.87v4.263a1 1 0 001.555.832l3.197-2.132a1 1 0 000-1.664z"
+              ></path>
+              <path
+                stroke-linecap="round"
+                stroke-linejoin="round"
+                stroke-width="2"
+                d="M21 12a9 9 0 11-18 0 9 9 0 0118 0z"
+              ></path>
+            </svg>
+          </div>
+          {% endif %}
+          <div class="p-4">
+            <p class="font-bold text-sm truncate">{{ video.prompt }}</p>
+            <p class="text-xs text-gray-400 mt-1">
+              Video ID: {{ video.id[:8] }}...
+            </p>
+            <p class="text-xs text-gray-500 mt-1">
+              {{ video.created_at | fromisoformat | humantime }}
+            </p>
+          </div>
+        </a>
+        {% endfor %}
+      </div>
+      {% else %}
+      <p class="text-gray-400">
+        You haven't generated any videos yet.
+        <a
+          href="{{ url_for('generate_video') }}"
+          class="text-blue-400 hover:underline"
+          >Generate one now</a
+        >.
+      </p>
+      {% endif %}
+    </div>
+
+    <!-- Uploaded Images -->
+    <div>
+      <h2 class="text-2xl font-semibold mb-4 border-b border-gray-700 pb-2">
+        My Uploads
+      </h2>
+      {% if uploads %}
+      <div
+        class="grid grid-cols-1 md:grid-cols-2 lg:grid-cols-3 xl:grid-cols-4 gap-6"
+      >
+        {% for image in uploads %}
+        <a
+          href="{{ url_for('upload_detail', image_id=image.id) }}"
+          class="block bg-gray-800 rounded-lg shadow-lg overflow-hidden hover:shadow-2xl transition-shadow duration-300"
+        >
+          <img
+            src="{{ url_for('serve_uploaded_image', image_id=image.id) }}"
+            alt="{{ image.filename }}"
+            class="w-full h-48 object-cover"
+          />
+          <div class="p-4">
+            <p class="font-bold text-sm truncate">{{ image.filename }}</p>
+            <p class="text-xs text-gray-400 mt-1">
+              Upload ID: {{ image.id[:8] }}...
+            </p>
+            <p class="text-xs text-gray-500 mt-1">
+              {{ image.uploaded_at | fromisoformat | humantime }}
+            </p>
+          </div>
+        </a>
+        {% endfor %}
+      </div>
+      {% else %}
+      <p class="text-gray-400">You haven't uploaded any images.</p>
+      {% endif %}
+    </div>
+  </div>
+
+  <!-- Infinite Scroll Loading Indicator -->
+  <div id="loading-indicator" class="flex justify-center py-8 hidden">
+    <div class="spinner"></div>
+  </div>
+  {% endblock %} {% block scripts %}
+  <script>
+    document.addEventListener("DOMContentLoaded", function () {
+      const galleryContainers = document.querySelectorAll(".grid[data-grid]");
+      const loadingIndicator = document.getElementById("loading-indicator");
+      const container = document.querySelector(".container[data-current-page]");
+      const currentPage = parseInt(container.dataset.currentPage);
+      const perPage = parseInt(container.dataset.perPage);
+      let isLoading = false;
+      let hasMore = true;
+
+      // Add data-grid attribute to all gallery grids
+      document
+        .querySelectorAll(".grid")
+        .forEach((grid) => grid.setAttribute("data-grid", ""));
+
+      // Infinite scroll handler
+      window.addEventListener("scroll", async function () {
+        if (!hasMore || isLoading) return;
+
+        const scrollPosition = window.innerHeight + window.scrollY;
+        const bottomThreshold = document.body.offsetHeight - 1000;
+
+        if (scrollPosition >= bottomThreshold) {
+          isLoading = true;
+          loadingIndicator.classList.remove("hidden");
+          // TODO: Implement actual fetching of next page of results and appending to the correct grid(s)
+          // For demo purposes, we'll just simulate a delay and then hide the loading indicator
+          // Simulate API call for next page
+          // In real implementation, replace with actual backend fetch
+          setTimeout(() => {
+            isLoading = false;
+            loadingIndicator.classList.add("hidden");
+            // Real app would fetch /generate/images?page=${currentPage +1}&limit=${perPage}
+            // and /generate/videos similarly
+          }, 1500);
+        }
+      });
+      // Cancel pending video buttons
+      document.querySelectorAll(".cancel-pending-btn").forEach((btn) => {
+        btn.addEventListener("click", async (e) => {
+          e.preventDefault();
+          e.stopPropagation();
+          const videoId = btn.dataset.videoId;
+          const msgEl = btn.parentElement.querySelector(".cancel-pending-msg");
+          btn.disabled = true;
+          btn.textContent = "Cancelling…";
+          try {
+            const resp = await fetch(
+              "/generate/video/" + encodeURIComponent(videoId) + "/cancel",
+              { method: "POST" },
+            );
+            if (resp.ok) {
+              btn.classList.add("hidden");
+              if (msgEl) {
+                msgEl.textContent = "Cancelled";
+                msgEl.classList.remove("hidden", "text-red-500");
+                msgEl.classList.add("text-gray-300");
+              }
+              const card = document.querySelector(
+                '[data-pending-video-id="' + videoId + '"]',
+              );
+              if (card) {
+                const statusSpan = card.querySelector(".text-yellow-400");
+                if (statusSpan) {
+                  statusSpan.textContent = "cancelled";
+                  statusSpan.classList.remove("text-yellow-400");
+                  statusSpan.classList.add("text-gray-400");
+                }
+              }
+            } else {
+              const data = await resp.json().catch(() => ({}));
+              btn.disabled = false;
+              btn.textContent = "Cancel";
+              if (msgEl) {
+                msgEl.textContent = data.detail || "Failed";
+                msgEl.classList.remove("hidden");
+                msgEl.classList.add("text-red-500");
+              }
+            }
+          } catch (err) {
+            btn.disabled = false;
+            btn.textContent = "Cancel";
+            if (msgEl) {
+              msgEl.textContent = "Error";
+              msgEl.classList.remove("hidden");
+              msgEl.classList.add("text-red-500");
+            }
+          }
+        });
+      });
+    });
+  </script>
+  {% endblock %}
+</div>

frontend/app/templates/generate.html

@@ -1,5 +1,5 @@
-{% extends "base.html" %} {% block title %}Generate — AI Allucanget{% endblock
-%} {% block content %}
+{% extends "base.html" %} {% block title %}Generate — All You Can GET AI{%
+endblock %} {% block content %}
 <div class="card">
   <h1>Generate</h1>
   <p class="text-muted">

frontend/app/templates/generate_image.html

@@ -1,53 +1,62 @@
 {% extends "base.html" %}
-{% block title %}Image Generation — AI Allucanget{% endblock %}
+{% block title %}Image Generation — All You Can GET AI{% endblock %}
 {% block content %}
 <div class="card">
   <h1>Image Generation</h1>
-  <form method="post">
+  <form method="post" enctype="multipart/form-data">
     <label for="model">Model</label>
+    {% if models %}
+    <select id="model" name="model" required>
+      {% for m in models %}
+      <option value="{{ m.id }}" {{ "selected" if request.form.get('model', '') == m.id else "" }}>{{ m.name }}</option>
+      {% endfor %}
+    </select>
+    {% else %}
     <input id="model" name="model" type="text" required
-           placeholder="e.g. openai/dall-e-3"
+           placeholder="e.g. google/gemini-2.5-flash-image"
            value="{{ request.form.get('model', '') }}">
+    <p class="text-muted mt-1">No models available</p>
+    {% endif %}
 
     <label for="prompt">Prompt</label>
     <textarea id="prompt" name="prompt" rows="4" required
               placeholder="Describe the image you want…">{{ request.form.get('prompt', '') }}</textarea>
 
-    <label for="size">Size</label>
-    <select id="size" name="size">
-      <option value="1024x1024" {% if request.form.get('size','1024x1024')=='1024x1024' %}selected{% endif %}>1024×1024</option>
-      <option value="1792x1024" {% if request.form.get('size')=='1792x1024' %}selected{% endif %}>1792×1024 (landscape)</option>
-      <option value="1024x1792" {% if request.form.get('size')=='1024x1792' %}selected{% endif %}>1024×1792 (portrait)</option>
-      <option value="512x512"   {% if request.form.get('size')=='512x512' %}selected{% endif %}>512×512</option>
-    </select>
-
     <label for="aspect_ratio">Aspect ratio</label>
     <select id="aspect_ratio" name="aspect_ratio">
-      <option value="">Auto (default)</option>
-      <option value="1:1" {% if request.form.get('aspect_ratio')=='1:1' %}selected{% endif %}>1:1 (square)</option>
-      <option value="16:9" {% if request.form.get('aspect_ratio')=='16:9' %}selected{% endif %}>16:9 (landscape)</option>
-      <option value="9:16" {% if request.form.get('aspect_ratio')=='9:16' %}selected{% endif %}>9:16 (portrait)</option>
-      <option value="4:3" {% if request.form.get('aspect_ratio')=='4:3' %}selected{% endif %}>4:3</option>
-      <option value="3:4" {% if request.form.get('aspect_ratio')=='3:4' %}selected{% endif %}>3:4</option>
-      <option value="3:2" {% if request.form.get('aspect_ratio')=='3:2' %}selected{% endif %}>3:2</option>
-      <option value="2:3" {% if request.form.get('aspect_ratio')=='2:3' %}selected{% endif %}>2:3</option>
+      <option value="">Auto (default 1:1)</option>
+      <option value="1:1" {{ "selected" if request.form.get('aspect_ratio')=='1:1' else "" }}>1:1 (square)</option>
+      <option value="16:9" {{ "selected" if request.form.get('aspect_ratio')=='16:9' else "" }}>16:9 (landscape)</option>
+      <option value="9:16" {{ "selected" if request.form.get('aspect_ratio')=='9:16' else "" }}>9:16 (portrait)</option>
+      <option value="4:3" {{ "selected" if request.form.get('aspect_ratio')=='4:3' else "" }}>4:3</option>
+      <option value="3:4" {{ "selected" if request.form.get('aspect_ratio')=='3:4' else "" }}>3:4</option>
+      <option value="3:2" {{ "selected" if request.form.get('aspect_ratio')=='3:2' else "" }}>3:2</option>
+      <option value="2:3" {{ "selected" if request.form.get('aspect_ratio')=='2:3' else "" }}>2:3</option>
     </select>
 
     <label for="image_size">Resolution</label>
     <select id="image_size" name="image_size">
-      <option value="">Auto (default)</option>
-      <option value="0.5K" {% if request.form.get('image_size')=='0.5K' %}selected{% endif %}>0.5K (low)</option>
-      <option value="1K" {% if request.form.get('image_size')=='1K' %}selected{% endif %}>1K (standard)</option>
-      <option value="2K" {% if request.form.get('image_size')=='2K' %}selected{% endif %}>2K (high)</option>
-      <option value="4K" {% if request.form.get('image_size')=='4K' %}selected{% endif %}>4K (ultra)</option>
+      <option value="">Auto (default 1K)</option>
+      <option value="0.5K" {{ "selected" if request.form.get('image_size')=='0.5K' else "" }}>0.5K (low)</option>
+      <option value="1K" {{ "selected" if request.form.get('image_size')=='1K' else "" }}>1K (standard)</option>
+      <option value="2K" {{ "selected" if request.form.get('image_size')=='2K' else "" }}>2K (high)</option>
+      <option value="4K" {{ "selected" if request.form.get('image_size')=='4K' else "" }}>4K (ultra)</option>
     </select>
 
-    <label for="n">Number of images</label>
-    <select id="n" name="n">
-      <option value="1" {% if request.form.get('n','1')=='1' %}selected{% endif %}>1</option>
-      <option value="2" {% if request.form.get('n')=='2' %}selected{% endif %}>2</option>
-      <option value="4" {% if request.form.get('n')=='4' %}selected{% endif %}>4</option>
-    </select>
+    <label for="reference_image">Reference image (optional)</label>
+    <input
+      id="reference_image"
+      name="reference_image"
+      type="file"
+      accept="image/png,image/jpeg,image/webp,image/gif"
+    >
+    <p class="text-muted mt-1" id="reference-image-help">
+      Upload an image to use as visual reference (image-to-image).
+    </p>
+    <div class="image-upload-preview" id="image-upload-preview" hidden>
+      <p class="text-muted" id="image-upload-filename"></p>
+      <img id="image-upload-preview-img" alt="Uploaded reference image preview" class="generated-image">
+    </div>
 
     <button type="submit">Generate image</button>
   </form>
@@ -60,7 +69,9 @@
     <div class="result">
       <h2>Generated image{{ 's' if result.images|length > 1 }}</h2>
       {% for img in result.images %}
+        {% if img.url %}
           <img src="{{ img.url }}" alt="Generated image" class="generated-image">
+        {% endif %}
         {% if img.revised_prompt %}
           <p class="text-muted mt-1" style="font-size:0.8rem;">{{ img.revised_prompt }}</p>
         {% endif %}

frontend/app/templates/generate_text.html

@@ -1,44 +1,93 @@
-{% extends "base.html" %} {% block title %}Text Generation — AI Allucanget{%
-endblock %} {% block content %}
-<div class="card">
-  <h1>Text Generation</h1>
-  <form method="post">
-    <label for="model">Model</label>
+{% extends "base.html" %} {% block title %}Text Generation — All You Can GET
+AI{% endblock %} {% block content %}
+<div class="card chat-page">
+  <div class="chat-header">
+    <h1>Text Chat</h1>
+    <form method="post" style="display: inline">
+      <input type="hidden" name="action" value="clear" />
+      <button type="submit" class="btn-secondary btn-sm">New Chat</button>
+    </form>
+  </div>
+
+  <!-- Config row -->
+  <details class="chat-config" {% if not chat_history %}open{% endif %}>
+    <summary>Model &amp; System Prompt</summary>
+    <div class="chat-config-body">
+      <label for="cfg-model">Model</label>
+      {% if models %}
+      <select id="cfg-model" form="chat-form" name="model" required>
+        {% for m in models %}
+        <option value="{{ m.id }}" {{ "selected" if current_model == m.id else "" }}>{{ m.name }}</option>
+        {% endfor %}
+      </select>
+      {% else %}
       <input
-      id="model"
+        id="cfg-model"
+        form="chat-form"
         name="model"
         type="text"
         required
         placeholder="e.g. openai/gpt-4o"
-      value="{{ request.form.get('model', '') }}"
+        value="{{ current_model }}"
       />
+      <p class="text-muted mt-1">No models available</p>
+      {% endif %}
 
-    <label for="prompt">Prompt</label>
+      <label for="cfg-sys">System prompt (optional)</label>
       <textarea
-      id="prompt"
-      name="prompt"
-      rows="5"
-      required
-      placeholder="Describe what you want…"
+        id="cfg-sys"
+        form="chat-form"
+        name="system_prompt"
+        rows="2"
+        placeholder="Set behavior/instructions for assistant…"
       >
-{{ request.form.get('prompt', '') }}</textarea
+{{ system_prompt }}</textarea
       >
+    </div>
+  </details>
 
-    <button type="submit">Generate text</button>
-  </form>
-
-  {% if error %}
-  <div class="alert alert-error mt-2">{{ error }}</div>
-  {% endif %} {% if result %}
-  <div class="result">
-    <h2>Result</h2>
-    <pre>{{ result.content }}</pre>
-    {% if result.usage %}
-    <p class="text-muted mt-1" style="font-size: 0.8rem">
-      Tokens: {{ result.usage.get('total_tokens', '—') }}
-    </p>
+  <!-- Chat history -->
+  <div class="chat-history" id="chat-history">
+    {% if not chat_history %}
+    <p class="chat-empty">No messages yet. Start the conversation below.</p>
+    {% endif %} {% for msg in chat_history %} {% if msg.role == "user" %}
+    <div class="chat-bubble chat-bubble--user">
+      <span class="bubble-role">You</span>
+      <div class="bubble-content">{{ msg.content }}</div>
+    </div>
+    {% elif msg.role == "assistant" %}
+    <div class="chat-bubble chat-bubble--assistant">
+      <span class="bubble-role">Assistant</span>
+      <div class="bubble-content">{{ msg.content }}</div>
+      {% if msg.usage %}
+      <span class="bubble-meta"
+        >{{ msg.usage.get('total_tokens', '') }} tokens</span
+      >
       {% endif %}
     </div>
+    {% endif %} {% endfor %} {% if error %}
+    <div class="alert alert-error">{{ error }}</div>
     {% endif %}
+  </div>
+
+  <!-- Input -->
+  <form id="chat-form" method="post" class="chat-input-row">
+    <input type="hidden" name="action" value="send" />
+    <textarea
+      name="prompt"
+      id="prompt"
+      rows="2"
+      required
+      placeholder="Type a message…"
+      class="chat-input-textarea"
+    ></textarea>
+    <button type="submit" class="btn-primary">Send</button>
+  </form>
 </div>
+
+<script>
+  // Auto-scroll chat to bottom
+  const hist = document.getElementById("chat-history");
+  if (hist) hist.scrollTop = hist.scrollHeight;
+</script>
 {% endblock %}

frontend/app/templates/generate_video.html

@@ -1,5 +1,5 @@
-{% extends "base.html" %} {% block title %}Video Generation — AI Allucanget{%
-endblock %} {% block content %}
+{% extends "base.html" %} {% block title %}Video Generation — All You Can GET
+AI{% endblock %} {% block content %}
 <div class="card">
   <h1>Video Generation</h1>
 
@@ -19,6 +19,13 @@ endblock %} {% block content %}
         <input type="hidden" name="mode" value="text" />
 
         <label for="model-t">Model</label>
+        {% if models %}
+        <select id="model-t" name="model" required>
+          {% for m in models %}
+          <option value="{{ m.id }}" {% if request.form.get('model', '') == m.id and request.form.get('mode','text')=='text' %}selected{% endif %}>{{ m.name }}</option>
+          {% endfor %}
+        </select>
+        {% else %}
         <input
           id="model-t"
           name="model"
@@ -27,6 +34,8 @@ endblock %} {% block content %}
           placeholder="e.g. openai/sora-2-pro"
           value="{{ request.form.get('model', '') if request.form.get('mode','text')=='text' else '' }}"
         />
+        <p class="text-muted mt-1">No models available</p>
+        {% endif %}
 
         <label for="prompt-t">Prompt</label>
         <textarea
@@ -54,21 +63,14 @@ endblock %} {% block content %}
           <option value="1080p">1080p</option>
         </select>
 
-        <label for="duration-t"
-          >Duration: <span id="duration-t-val">5</span>s</label
-        >
-        <input
-          type="range"
-          id="duration-t"
-          name="duration_seconds"
-          min="5"
-          max="60"
-          step="1"
-          value="5"
-          oninput="
-            document.getElementById('duration-t-val').textContent = this.value
-          "
-        />
+        <label for="duration-t">Duration (seconds)</label>
+        <select id="duration-t" name="duration_seconds">
+          <option value="4">4s</option>
+          <option value="8">8s</option>
+          <option value="12" selected>12s</option>
+          <option value="16">16s</option>
+          <option value="20">20s</option>
+        </select>
 
         <button type="submit">Generate video</button>
       </form>
@@ -80,6 +82,13 @@ endblock %} {% block content %}
         <input type="hidden" name="mode" value="image" />
 
         <label for="model-i">Model</label>
+        {% if models %}
+        <select id="model-i" name="model" required>
+          {% for m in models %}
+          <option value="{{ m.id }}" {% if request.form.get('model', '') == m.id and request.form.get('mode')=='image' %}selected{% endif %}>{{ m.name }}</option>
+          {% endfor %}
+        </select>
+        {% else %}
         <input
           id="model-i"
           name="model"
@@ -88,6 +97,8 @@ endblock %} {% block content %}
           placeholder="e.g. openai/sora-2-pro"
           value="{{ request.form.get('model', '') if request.form.get('mode')=='image' else '' }}"
         />
+        <p class="text-muted mt-1">No models available</p>
+        {% endif %}
 
         <label for="image_url">Source image URL</label>
         <input
@@ -125,21 +136,14 @@ endblock %} {% block content %}
           <option value="1080p">1080p</option>
         </select>
 
-        <label for="duration-i"
-          >Duration: <span id="duration-i-val">5</span>s</label
-        >
-        <input
-          type="range"
-          id="duration-i"
-          name="duration_seconds"
-          min="5"
-          max="60"
-          step="1"
-          value="5"
-          oninput="
-            document.getElementById('duration-i-val').textContent = this.value
-          "
-        />
+        <label for="duration-i">Duration (seconds)</label>
+        <select id="duration-i" name="duration_seconds">
+          <option value="4">4s</option>
+          <option value="8">8s</option>
+          <option value="12" selected>12s</option>
+          <option value="16">16s</option>
+          <option value="20">20s</option>
+        </select>
 
         <button type="submit">Generate video from image</button>
       </form>
@@ -151,9 +155,9 @@ endblock %} {% block content %}
   {% endif %} {% if result %}
   <div class="result">
     <h2>Video job</h2>
-    <p>Job ID: <code>{{ result.id }}</code></p>
-    {% if result.status in ('queued', 'processing') and result.polling_url %}
-    <div id="video-poll-status" data-polling-url="{{ result.polling_url }}">
+    <p>Job ID: <code>{{ result.db_id or result.id }}</code></p>
+    {% if result.status in ('queued', 'processing') and result.db_id %}
+    <div id="video-poll-status" data-video-id="{{ result.db_id }}">
       <p>
         <span id="poll-status-text"
           >Status: <strong>{{ result.status }}</strong></span
@@ -161,6 +165,13 @@ endblock %} {% block content %}
         &mdash; checking for updates every 5 s&hellip;
       </p>
       <div id="poll-video-container"></div>
+      <button
+        id="cancel-video-btn"
+        class="mt-2 px-4 py-2 bg-red-600 hover:bg-red-700 text-white rounded-md text-sm"
+      >
+        Cancel Job
+      </button>
+      <p id="cancel-msg" class="text-sm mt-2 hidden"></p>
     </div>
     {% elif result.video_url %}
     <video

frontend/app/templates/image_detail.html

@@ -0,0 +1,35 @@
+{% extends "base.html" %} {% block title %}Generated Image{% endblock %} {%
+block content %}
+<div class="container mx-auto px-4 py-8">
+  <a
+    href="{{ url_for('gallery') }}"
+    class="text-blue-400 hover:underline mb-4 inline-block"
+    >&larr; Back to Gallery</a
+  >
+
+  {% if image %}
+  <h1 class="text-2xl font-bold mb-4">Generated Image</h1>
+  <div class="bg-gray-800 rounded-lg shadow-lg overflow-hidden">
+    <img
+      src="{{ image.image_data }}"
+      alt="{{ image.prompt }}"
+      class="w-full object-contain"
+    />
+    <div class="p-6">
+      <h2 class="text-xl font-semibold mb-2">Prompt</h2>
+      <p class="text-gray-300 bg-gray-900 p-3 rounded-md">{{ image.prompt }}</p>
+      <div class="mt-4 text-sm text-gray-400">
+        <p><strong>Model:</strong> {{ image.model_id }}</p>
+        <p>
+          <strong>Created:</strong> {{ image.created_at | fromisoformat |
+          humantime }}
+        </p>
+      </div>
+    </div>
+  </div>
+  {% else %}
+  <h1 class="text-2xl font-bold">Image not found</h1>
+  <p class="text-gray-400 mt-2">Could not find details for this image.</p>
+  {% endif %}
+</div>
+{% endblock %}

frontend/app/templates/login.html

@@ -1,14 +1,13 @@
-{% extends "base.html" %}
-{% block title %}Log in — AI Allucanget{% endblock %}
-{% block content %}
+{% extends "base.html" %} {% block title %}Log in — All You Can GET AI{%
+endblock %} {% block content %}
 <div class="card">
   <h1>Log in</h1>
   <form method="post">
     <label for="email">Email</label>
-    <input id="email" name="email" type="email" required autofocus>
+    <input id="email" name="email" type="email" required autofocus />
 
     <label for="password">Password</label>
-    <input id="password" name="password" type="password" required>
+    <input id="password" name="password" type="password" required />
 
     <button type="submit">Log in</button>
   </form>

frontend/app/templates/profile.html

@@ -1,5 +1,5 @@
-{% extends "base.html" %} {% block title %}Profile — AI Allucanget{% endblock %}
-{% block content %}
+{% extends "base.html" %} {% block title %}Profile — All You Can GET AI{%
+endblock %} {% block content %}
 <div class="card">
   <h1>Your Profile</h1>
 

frontend/app/templates/register.html

@@ -1,14 +1,19 @@
-{% extends "base.html" %}
-{% block title %}Register — AI Allucanget{% endblock %}
-{% block content %}
+{% extends "base.html" %} {% block title %}Register — All You Can GET AI{%
+endblock %} {% block content %}
 <div class="card">
   <h1>Create account</h1>
   <form method="post">
     <label for="email">Email</label>
-    <input id="email" name="email" type="email" required autofocus>
+    <input id="email" name="email" type="email" required autofocus />
 
     <label for="password">Password</label>
-    <input id="password" name="password" type="password" required minlength="8">
+    <input
+      id="password"
+      name="password"
+      type="password"
+      required
+      minlength="8"
+    />
 
     <button type="submit">Register</button>
   </form>

frontend/app/templates/upload_detail.html

@@ -0,0 +1,40 @@
+{% extends "base.html" %} {% block title %}Uploaded Image{% endblock %} {% block
+content %}
+<div class="container mx-auto px-4 py-8">
+  <a
+    href="{{ url_for('gallery') }}"
+    class="text-blue-400 hover:underline mb-4 inline-block"
+    >&larr; Back to Gallery</a
+  >
+
+  {% if image %}
+  <h1 class="text-2xl font-bold mb-4">Uploaded Image</h1>
+  <div class="bg-gray-800 rounded-lg shadow-lg overflow-hidden">
+    <img
+      src="{{ url_for('serve_uploaded_image', image_id=image.id) }}"
+      alt="{{ image.filename }}"
+      class="w-full object-contain"
+    />
+    <div class="p-6">
+      <h2 class="text-xl font-semibold mb-2">Details</h2>
+      <div class="mt-4 text-sm text-gray-400">
+        <p><strong>Filename:</strong> {{ image.filename }}</p>
+        <p><strong>Content Type:</strong> {{ image.content_type }}</p>
+        <p>
+          <strong>Size:</strong> {{ (image.size_bytes / 1024) | round(2) }} KB
+        </p>
+        <p>
+          <strong>Uploaded:</strong> {{ image.created_at | fromisoformat |
+          humantime }}
+        </p>
+      </div>
+    </div>
+  </div>
+  {% else %}
+  <h1 class="text-2xl font-bold">Image not found</h1>
+  <p class="text-gray-400 mt-2">
+    Could not find details for this uploaded image.
+  </p>
+  {% endif %}
+</div>
+{% endblock %}

frontend/app/templates/video_detail.html

@@ -0,0 +1,77 @@
+{% extends "base.html" %} {% block title %}Generated Video{% endblock %} {%
+block content %}
+<div class="container mx-auto px-4 py-8">
+  <a
+    href="{{ url_for('gallery') }}"
+    class="text-blue-400 hover:underline mb-4 inline-block"
+    >&larr; Back to Gallery</a
+  >
+
+  {% if video %}
+  <h1 class="text-2xl font-bold mb-4">Video Generation Job</h1>
+  <div class="bg-gray-800 rounded-lg shadow-lg overflow-hidden">
+    {% if video.status == 'completed' and video.video_url %}
+    <video src="{{ video.video_url }}" controls class="w-full"></video>
+    {% elif video.status in ('queued', 'processing') %}
+    <div
+      class="w-full bg-black aspect-video flex flex-col items-center justify-center p-6 text-center"
+      id="video-poll-status"
+      data-video-id="{{ video.id }}"
+    >
+      <p class="text-xl font-semibold">
+        Status: <strong id="poll-status-text">{{ video.status }}</strong>
+      </p>
+      <p class="text-gray-400 mt-2">
+        Your video is being processed. This page will update automatically when
+        it's ready.
+      </p>
+      <div class="spinner mt-4"></div>
+      <button
+        id="cancel-video-btn"
+        class="mt-4 px-4 py-2 bg-red-600 hover:bg-red-700 text-white rounded-md text-sm"
+      >
+        Cancel Job
+      </button>
+      <p id="cancel-msg" class="text-sm mt-2 hidden"></p>
+    </div>
+    {% elif video.status == 'failed' %}
+    <div
+      class="w-full bg-black aspect-video flex flex-col items-center justify-center p-6 text-center"
+    >
+      <p class="text-xl font-semibold text-red-500">Generation Failed</p>
+      <p class="text-gray-400 mt-2">
+        {{ video.error or 'An unknown error occurred.' }}
+      </p>
+    </div>
+    {% else %}
+    <div
+      class="w-full bg-black aspect-video flex flex-col items-center justify-center p-6 text-center"
+    >
+      <p class="text-xl font-semibold">Video Not Available</p>
+      <p class="text-gray-400 mt-2">Status: {{ video.status }}</p>
+    </div>
+    {% endif %}
+
+    <div class="p-6">
+      <h2 class="text-xl font-semibold mb-2">Prompt</h2>
+      <p class="text-gray-300 bg-gray-900 p-3 rounded-md">{{ video.prompt }}</p>
+      <div class="mt-4 text-sm text-gray-400">
+        <p><strong>Model:</strong> {{ video.model_id }}</p>
+        <p><strong>Job ID:</strong> <code>{{ video.job_id }}</code></p>
+        <p>
+          <strong>Created:</strong> {{ video.created_at | fromisoformat |
+          humantime }}
+        </p>
+        <p>
+          <strong>Last Update:</strong> {{ video.updated_at | fromisoformat |
+          humantime }}
+        </p>
+      </div>
+    </div>
+  </div>
+  {% else %}
+  <h1 class="text-2xl font-bold">Video job not found</h1>
+  <p class="text-gray-400 mt-2">Could not find details for this video job.</p>
+  {% endif %}
+</div>
+{% endblock %}

frontend/tests/test_frontend.py

@@ -148,9 +148,12 @@ def test_dashboard_requires_login(client):
 
 def test_dashboard_renders_user_info(client):
     _set_auth(client)
-    mock = _mock_response(
+    me_mock = _mock_response(
         200, {"id": "1", "email": "u@example.com", "role": "user"})
-    with patch("frontend.app.main.httpx.request", return_value=mock):
+    images_mock = _mock_response(200, [])
+    gen_images_mock = _mock_response(200, [])
+    gen_videos_mock = _mock_response(200, [])
+    with patch("frontend.app.main.httpx.request", side_effect=[me_mock, images_mock, gen_images_mock, gen_videos_mock]):
         resp = client.get("/dashboard")
     assert resp.status_code == 200
     assert b"u@example.com" in resp.data
@@ -171,7 +174,7 @@ def test_generate_text_page_renders(client):
     _set_auth(client)
     resp = client.get("/generate/text")
     assert resp.status_code == 200
-    assert b"Text Generation" in resp.data
+    assert b"Text Chat" in resp.data
 
 
 def test_generate_text_requires_login(client):
@@ -182,13 +185,108 @@ def test_generate_text_requires_login(client):
 
 def test_generate_text_success(client):
     _set_auth(client)
-    mock = _mock_response(
+    gen_mock = _mock_response(
         200, {"id": "g1", "model": "openai/gpt-4o", "content": "Hello world", "usage": None})
-    with patch("frontend.app.main.httpx.request", return_value=mock):
+    models_mock = _mock_response(200, [
+        {"id": "openai/gpt-4o", "name": "GPT-4o", "modality": "text"}
+    ])
+    with patch("frontend.app.main.httpx.request", side_effect=[gen_mock, models_mock]):
         resp = client.post(
-            "/generate/text", data={"model": "openai/gpt-4o", "prompt": "Say hello"})
+            "/generate/text",
+            data={"model": "openai/gpt-4o", "prompt": "Say hello", "action": "send"})
     assert resp.status_code == 200
     assert b"Hello world" in resp.data
+    assert b"chat-bubble--assistant" in resp.data
+
+
+def test_generate_text_page_shows_optional_system_prompt(client):
+    _set_auth(client)
+    models_mock = _mock_response(200, [])
+    with patch("frontend.app.main.httpx.request", return_value=models_mock):
+        resp = client.get("/generate/text")
+    assert resp.status_code == 200
+    assert b"System prompt (optional)" in resp.data
+    assert b'name="system_prompt"' in resp.data
+
+
+def test_generate_text_forwards_system_prompt(client):
+    _set_auth(client)
+    gen_mock = _mock_response(
+        200, {"id": "g1", "model": "openai/gpt-4o", "content": "Hello world", "usage": None})
+    models_mock = _mock_response(200, [
+        {"id": "openai/gpt-4o", "name": "GPT-4o", "modality": "text"}
+    ])
+
+    with patch("frontend.app.main.httpx.request", side_effect=[gen_mock, models_mock]) as mock_request:
+        resp = client.post(
+            "/generate/text",
+            data={
+                "model": "openai/gpt-4o",
+                "prompt": "Say hello",
+                "system_prompt": "You are concise.",
+                "action": "send",
+            },
+        )
+
+    assert resp.status_code == 200
+    first_call_kwargs = mock_request.call_args_list[0].kwargs
+    assert first_call_kwargs["json"]["system_prompt"] == "You are concise."
+    # Messages array sent (not bare prompt)
+    assert "messages" in first_call_kwargs["json"]
+
+
+def test_generate_text_chat_history_accumulates(client):
+    """Second message includes prior user+assistant turns in messages array."""
+    _set_auth(client)
+
+    turn1_gen = _mock_response(
+        200, {"id": "g1", "model": "openai/gpt-4o", "content": "Turn 1 reply", "usage": None})
+    turn1_models = _mock_response(
+        200, [{"id": "openai/gpt-4o", "name": "GPT-4o", "modality": "text"}])
+    turn2_gen = _mock_response(
+        200, {"id": "g2", "model": "openai/gpt-4o", "content": "Turn 2 reply", "usage": None})
+    turn2_models = _mock_response(
+        200, [{"id": "openai/gpt-4o", "name": "GPT-4o", "modality": "text"}])
+
+    with patch("frontend.app.main.httpx.request", side_effect=[turn1_gen, turn1_models]):
+        client.post(
+            "/generate/text", data={"model": "openai/gpt-4o", "prompt": "First", "action": "send"})
+
+    with patch("frontend.app.main.httpx.request", side_effect=[turn2_gen, turn2_models]) as mock_req:
+        resp = client.post(
+            "/generate/text", data={"model": "openai/gpt-4o", "prompt": "Second", "action": "send"})
+
+    assert resp.status_code == 200
+    assert b"Turn 1 reply" in resp.data
+    assert b"Turn 2 reply" in resp.data
+    # Backend received 3 messages: First(user), Turn1(assistant), Second(user)
+    sent_messages = mock_req.call_args_list[0].kwargs["json"]["messages"]
+    assert len(sent_messages) == 3
+    assert sent_messages[0]["role"] == "user" and sent_messages[0]["content"] == "First"
+    assert sent_messages[1]["role"] == "assistant"
+    assert sent_messages[2]["role"] == "user" and sent_messages[2]["content"] == "Second"
+
+
+def test_generate_text_clear_resets_history(client):
+    """Clear action removes session history and redirects."""
+    _set_auth(client)
+
+    gen_mock = _mock_response(
+        200, {"id": "g1", "model": "openai/gpt-4o", "content": "Reply", "usage": None})
+    models_mock = _mock_response(
+        200, [{"id": "openai/gpt-4o", "name": "GPT-4o", "modality": "text"}])
+    with patch("frontend.app.main.httpx.request", side_effect=[gen_mock, models_mock]):
+        client.post(
+            "/generate/text", data={"model": "openai/gpt-4o", "prompt": "Hi", "action": "send"})
+
+    clear_resp = client.post("/generate/text", data={"action": "clear"})
+    assert clear_resp.status_code == 302
+
+    models_mock2 = _mock_response(
+        200, [{"id": "openai/gpt-4o", "name": "GPT-4o", "modality": "text"}])
+    with patch("frontend.app.main.httpx.request", return_value=models_mock2):
+        get_resp = client.get("/generate/text")
+    assert b"No messages yet" in get_resp.data
 
 
 def test_generate_image_page_renders(client):
@@ -196,6 +294,7 @@ def test_generate_image_page_renders(client):
     resp = client.get("/generate/image")
     assert resp.status_code == 200
     assert b"Image Generation" in resp.data
+    assert b"reference_image" in resp.data
 
 
 def test_generate_image_success(client):
@@ -249,8 +348,11 @@ def test_generate_video_image_mode(client):
 
 def test_generate_upstream_error_shows_message(client):
     _set_auth(client)
-    mock = _mock_response(502, {"detail": "OpenRouter error: timeout"})
-    with patch("frontend.app.main.httpx.request", return_value=mock):
+    gen_mock = _mock_response(502, {"detail": "OpenRouter error: timeout"})
+    models_mock = _mock_response(200, [
+        {"id": "openai/gpt-4o", "name": "GPT-4o", "modality": "text"}
+    ])
+    with patch("frontend.app.main.httpx.request", side_effect=[gen_mock, models_mock]):
         resp = client.post(
             "/generate/text", data={"model": "openai/gpt-4o", "prompt": "Hi"})
     assert resp.status_code == 200
@@ -417,4 +519,104 @@ def test_video_generate_renders_polling_ui(client):
         })
     assert resp.status_code == 200
     assert b"video-poll-status" in resp.data
-    assert b"openrouter.ai/api/v1/videos/v1" in resp.data
+
+
+# ---------------------------------------------------------------------------
+# Image upload — frontend proxy + dashboard
+# ---------------------------------------------------------------------------
+
+def test_dashboard_shows_uploaded_images(client):
+    _set_auth(client)
+    me_mock = _mock_response(
+        200, {"id": "1", "email": "u@example.com", "role": "user"})
+    images_mock = _mock_response(200, [
+        {"id": "img-1", "filename": "cat.png", "content_type": "image/png",
+         "size_bytes": 1024, "created_at": "2026-04-29T10:00:00"},
+    ])
+    gen_images_mock = _mock_response(200, [])
+    gen_videos_mock = _mock_response(200, [])
+    with patch("frontend.app.main.httpx.request", side_effect=[me_mock, images_mock, gen_images_mock, gen_videos_mock]):
+        resp = client.get("/dashboard")
+    assert resp.status_code == 200
+    assert b"cat.png" in resp.data
+    assert b"img-1" in resp.data
+
+
+def test_dashboard_shows_generated_images(client):
+    _set_auth(client)
+    me_mock = _mock_response(
+        200, {"id": "1", "email": "u@example.com", "role": "user"})
+    images_mock = _mock_response(200, [])
+    gen_images_mock = _mock_response(200, [
+        {
+            "id": "gen-1",
+            "model_id": "google/gemini-2.5-flash-image",
+            "prompt": "A cat on the moon",
+            "image_data": "data:image/png;base64,abc123",
+            "created_at": "2026-04-29T10:00:00",
+        }
+    ])
+    gen_videos_mock = _mock_response(200, [])
+    with patch("frontend.app.main.httpx.request", side_effect=[me_mock, images_mock, gen_images_mock, gen_videos_mock]):
+        resp = client.get("/dashboard")
+    assert resp.status_code == 200
+    assert b"Generated images" in resp.data
+    assert b"A cat on the moon" in resp.data
+    assert b"data:image/png;base64,abc123" in resp.data
+
+
+def test_dashboard_no_images_section_when_empty(client):
+    _set_auth(client)
+    me_mock = _mock_response(
+        200, {"id": "1", "email": "u@example.com", "role": "user"})
+    images_mock = _mock_response(200, [])
+    gen_images_mock = _mock_response(200, [])
+    gen_videos_mock = _mock_response(200, [])
+    with patch("frontend.app.main.httpx.request", side_effect=[me_mock, images_mock, gen_images_mock, gen_videos_mock]):
+        resp = client.get("/dashboard")
+    assert resp.status_code == 200
+    assert b"Uploaded reference images" not in resp.data
+
+
+def test_serve_uploaded_image_proxy(client):
+    _set_auth(client)
+    img_bytes = b"\x89PNG\r\n\x1a\n"
+    mock = MagicMock()
+    mock.status_code = 200
+    mock.content = img_bytes
+    mock.headers = {"content-type": "image/png"}
+    with patch("frontend.app.main.httpx.request", return_value=mock):
+        resp = client.get("/images/img-1/file")
+    assert resp.status_code == 200
+    assert resp.content_type == "image/png"
+    assert resp.data == img_bytes
+
+
+def test_serve_uploaded_image_requires_login(client):
+    resp = client.get("/images/img-1/file")
+    assert resp.status_code == 302
+    assert "/login" in resp.headers["Location"]
+
+
+def test_serve_uploaded_image_not_found_proxied(client):
+    _set_auth(client)
+    mock = _mock_response(404, {"detail": "Image not found."})
+    mock.content = b""
+    with patch("frontend.app.main.httpx.request", return_value=mock):
+        resp = client.get("/images/bad-id/file")
+    assert resp.status_code == 404
+
+
+def test_generate_image_uploads_reference_then_generates(client):
+    _set_auth(client)
+    gen_mock = _mock_response(200, {
+        "id": "g2", "model": "openai/dall-e-3",
+        "images": [{"url": "https://example.com/out.png", "revised_prompt": None, "b64_json": None}]
+    })
+    # No file field → upload branch skipped; only generate call is made
+    with patch("frontend.app.main.httpx.request", return_value=gen_mock):
+        resp = client.post("/generate/image", data={
+            "model": "openai/dall-e-3", "prompt": "A cat", "n": "1", "size": "1024x1024",
+        }, content_type="multipart/form-data")
+    assert resp.status_code == 200
+    assert b"example.com/out.png" in resp.data

nginx/docker-compose.conf

@@ -1,53 +0,0 @@
-# Nginx reverse proxy configuration for Docker Compose deployment
-# Place this in docker-compose.yml volume mount to /etc/nginx/conf.d/default.conf
-
-# Backend API proxy - use Docker service name instead of localhost
-upstream backend {
-    server backend:12015;
-}
-
-# Frontend proxy - use Docker service name instead of localhost
-upstream frontend {
-    server frontend:12016;
-}
-
-server {
-    listen 80;
-    server_name _;
-
-    # Backend API proxy
-    location /api/ {
-        proxy_pass http://backend;
-        proxy_set_header Host $host;
-        proxy_set_header X-Real-IP $remote_addr;
-        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-        proxy_set_header X-Forwarded-Proto $scheme;
-
-        # WebSocket support (if needed in future)
-        proxy_http_version 1.1;
-        proxy_set_header Upgrade $http_upgrade;
-        proxy_set_header Connection "upgrade";
-    }
-
-    # Frontend proxy
-    location / {
-        proxy_pass http://frontend;
-        proxy_set_header Host $host;
-        proxy_set_header X-Real-IP $remote_addr;
-        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
-        proxy_set_header X-Forwarded-Proto $scheme;
-
-        # Static files caching
-        location ~* \.(css|js|png|jpg|jpeg|gif|ico|svg|woff|woff2|ttf|eot)$ {
-            proxy_pass http://frontend;
-            expires 30d;
-            add_header Cache-Control "public, immutable";
-        }
-    }
-
-    # Health check endpoint
-    location /health {
-        proxy_pass http://backend;
-        proxy_set_header Host $host;
-    }
-}

requirements.txt

@@ -1,5 +1,5 @@
 anyio
-bcrypt
+bcrypt==4.0.1
 blinker
 certifi
 cffi
@@ -20,7 +20,7 @@ itsdangerous
 Jinja2
 MarkupSafe
 packaging
-passlib
+passlib==1.7.4
 pluggy
 pyasn1
 pycparser