StepFunStep-Video
Step-Video T2V
Step-Video T2V is a 30B text-to-video model that can generate up to 204 frames. It uses a compressed video VAE and a diffusion transformer trained with flow matching.
Use the full video pipeline
Two text encoders cover Chinese and English prompts. The final training stage uses direct preference optimization to improve video output.StepFun stepvideo-t2v model card
Count frames separately from seconds
Practical guidance. The number of frames does not establish duration without a frame rate. Compare motion continuity and subject persistence across the full clip, not only a selected frame.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | stepvideo-t2vStep-Video-T2V |
| License | MIT.StepFun stepvideo-t2v model card |
| Compression | The VAE uses 16 by 16 spatial and 8-fold temporal compression.StepFun stepvideo-t2v model card |
Related models
- Step-Video TI2V
StepFun · Step-Video
A video-generation model that animates a starting image from a text instruction.
Launch date Downloadable - Step-Video T2V Turbo
StepFun · Step-Video
Step-Video T2V Turbo is the inference-step-distilled version of Step-Video T2V.
Launch date Downloadable