Wan Studio is in active development. Please don't run inference — every GPU click burns the maintainer's ZeroGPU quota. Follow along at github.com/techfreakworm/wan-studio.
Wan Studiostudio · video diffusion
Generation
Generate
Workspace
T2V — Text-to-Video
Resolution
0.55.1
02147483647
150
010
⌚ ZeroGPU reservation: ~110s
Send to:
I2V — Image-to-Video
Resolution
0.55.1
02147483647
150
010
⌚ ZeroGPU reservation: ~120s
Send to:
TI2V — Text+Image to Video (Wan 2.2-5B, via upstream wan package)
TI2V-5B is locked to 1280×704 (landscape) or 704×1280 (portrait), 121 frames @ 24 fps.
02147483647
150
010
⌚ ZeroGPU reservation: ~60s
Send to:
FLF2V — First-Last-Frame to Video (Wan 2.1 only)
Synthesize end frame from a prompt (T2I).
02147483647
150
010
⌚ ZeroGPU reservation: ~150s
Send to:
V2V — Video-to-Video Restyle
0.11
02147483647
150
010
⌚ ZeroGPU reservation: ~90s
Send to:
VACE — Versatile Animation Control & Editing (Wan 2.1 only)
02147483647
150
010
⌚ ZeroGPU reservation: ~180s
Send to:
S2V — Speech to Video (Wan 2.2, via upstream wan package)
Resolution
Duration: auto (driven by audio length)
02147483647
150
010
⌚ ZeroGPU reservation: ~240s (audio-driven)
Send to:
Animate — Character Animation & Replacement (Wan 2.2)
120
⚠ Pose+face preprocessing runs on CPU before GPU (~30s extra).
02147483647
150
010
⌚ ZeroGPU reservation: ~300s (xlarge tier)
Send to:
Gallery — session history
Generated videos will appear here.
Params will appear here.
Settings — Model Manager
Model load status appears here.
LoRA status appears here.
Wan Studio v0.1 · loading...
Run python -c 'from utils.backend import detect; print(detect())' to see the active backend.