Upload app.py

Upload app.py

Upload app.py

Upgrade to HF models: Qwen2.5-72B text, Qwen2.5-VL-7B vision, SDXL image. Remove Gemini dependency.

Fix env var loading: use function get_gemini_key() at runtime + add debug endpoints

Switch to Gemini + Pollinations AI (free, no HF token needed)

Fix API routing: static files AFTER API routes + CORS

Add FastAPI backend proxy