Diffusion Single File
MiniMax H3
comfyui
hyperflow-h3
studio-models-h3 / README.md
samisuteria's picture
Model card and NOTICE: the fl2va tower, link the loader
aba732d verified
|
Raw History Blame Contribute Delete
2.59 kB
---
license: other
license_name: minimax-h3-community-license-agreement
license_link: LICENSE
tags:
- comfyui
- diffusion-single-file
- hyperflow-h3
- minimax-h3
base_model:
- VideoRebirthAI/hyperflow-h3-flash
- MiniMaxAI/MiniMax-H3
---
# Studio models: MiniMax H3
Repackaged MiniMax H3 model files for [studio.no.dev](https://studio.no.dev), which runs them in ComfyUI and uncomfy.
Original model repositories:
- https://huggingface.co/VideoRebirthAI/hyperflow-h3-flash
- https://huggingface.co/MiniMaxAI/MiniMax-H3
## License
These files are Model Derivatives of MiniMax H3. Both the original model's license and MiniMax's H3 license apply: Video Rebirth distributes HyperFlow H3 Flash under the [MiniMax H3 Community License Agreement](LICENSE), and this repository redistributes it under the same Agreement with no additional terms. By downloading or using these files you accept that Agreement, including its territorial restrictions, commercial terms and Acceptable Use Policy. The [upstream LICENSE](https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE) governs.
[`NOTICE`](NOTICE) lists what changed from Video Rebirth's release. This repository is not an official MiniMax or Video Rebirth product. Powered by MiniMax H3.
## HyperFlow H3 Flash
Video Rebirth's two HyperFlow towers in ComfyUI's int8 layout. Both sample in 6 or 8 steps without PDD, and the int8 weights are copied unchanged from Video Rebirth's files.
| File | Tower | Use it for |
|---|---|---|
| `minimax_h3_fl2va_hyperflow_flash_int8_convrot_comfy.safetensors` (16.3 GB) | fl2va: 3 blocks dropped, 102 linears low-rank | Text to video, and image to video from a first frame or first and last frames (`MiniMaxH3ImageToVideo`) |
| `minimax_h3_ref2va_hyperflow_flash_int8_convrot_comfy.safetensors` (16.8 GB) | ref2va: 1 block dropped, 114 linears low-rank | Video from reference images and audio (`MiniMaxH3ReferenceToVideo`) |
They need [ComfyUI-Studio-HyperFlow](https://github.com/NoDevOrg/ComfyUI-Studio-HyperFlow)'s `HyperFlowInt8Loader` and Video Rebirth's [ComfyUI-HyperFlow-Flash](https://github.com/Video-Rebirth/ComfyUI-HyperFlow-Flash) for the schedule; ComfyUI's stock H3 loader won't load them. Use them with the stock MiniMax H3 text encoder and VAEs from [Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/MiniMax-H3).
```
πŸ“‚ ComfyUI/
β”œβ”€β”€ πŸ“‚ models/
β”‚ β”œβ”€β”€ πŸ“‚ diffusion_models/
β”‚ β”‚ β”œβ”€β”€ minimax_h3_fl2va_hyperflow_flash_int8_convrot_comfy.safetensors
β”‚ β”‚ └── minimax_h3_ref2va_hyperflow_flash_int8_convrot_comfy.safetensors
```