Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Right-size ZeroGPU duration requests (dynamic where workload-dependent)

Drop size=xlarge: the 14B fp8+AOTI pipeline fits default ZeroGPU size (matches the working i2v twin), halving quota cost

Wan 2.2 14B text-to-video (fp8 + Lightning LoRA + AOTI)