videolora training
Cogvideox 5B
Open-weight text-to-video model generating 10-second 720x480 clips with LoRA fine-tuning support.
Studio is in early access; availability resolves per project after sign-in.
- Endpoints
- 3
- Eligible
- 2
- Input schemas imported
- 3 / 3
- Delivery
- fal.ai
- Weights
- Not stated
- Developer
- Not stated in source
Overview
CogVideoX-5B generates 10-second videos from text prompts with temporal coherence. Full model weights and LoRA fine-tuning support give developers customizable open-source flexibility. Default output runs 720x480, configurable through a video-size parameter. Endpoints cover text-driven, image-driven, and video-to-video generation.
Capabilities
- 10-second text-to-video output
- Image-to-video and video-to-video endpoints
- Full weights plus LoRA fine-tuning
- Configurable 720x480 default resolution
Best for
- Marketing content creation
- Product demonstrations
- Social media video assets
Use cases
- Customizable open video pipelines
- Fine-tuned brand motion
- Short-form social clips
Endpoints
1 under review · 2 eligible. Cataloged is not the same as executable: Studio resolves which endpoints a project can run.
| Endpoint | Task | Catalog status | Inputs |
|---|---|---|---|
fal-ai/cogvideox-5b | unknown | Under review | 9 fields · requires prompt |
fal-ai/cogvideox-5b/image-to-video | image to video | Eligible | 10 fields · requires prompt, image_url |
fal-ai/cogvideox-5b/video-to-video | video to video | Eligible | 11 fields · requires prompt, video_url |
Questions
How long are clips?
10 seconds continuous with temporal coherence.
Can developers customize it?
Yes: full weights and LoRA fine-tuning support.