h3-models is an open-weight model from Xinyu. Its published files total 40.1 GB.
This repository preserves the H3 model files used by the owner's ComfyUI deployment. The models were copied from the owner's RunPod network volume without modifying the originals.
Model Card
This repository preserves the H3 model files used by the owner's ComfyUI deployment. The models were copied from the owner's RunPod network volume without modifying the originals. - unet/: H3 diffusion model - textencoders/: H3 Qwen3-VL text encoder - vae/: video and audio VAEs - loras/minimaxh3/: H3 LoRA collection - models-manifest.json: published only after all included files have verified sizes and SHA256 hashes For reproducible deployment, download the manifest and its listed model files from the same immutable repository commit. Verify every file against the manifest before making it available to ComfyUI. Preserve relative paths under the Pod's local model cache. The repository is…
Excerpt from the card by Xinyu.
Identity and Version
- Repository
- liuxin9494/h3-models
- Publisher
- Xinyu
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- Not stated by the source
- Languages
- Not stated by the source
- Revision
- 90d2bffb9201a5dcba221deebb336aa3bf16926b
- First published
- 2026-09-29
- Last updated
- 2026-09-30
Files and Weights
17 files, 40.1 GB in total. The weights are 14 files totalling 40.1 GB in safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| loras/minimaxh3/Blacked_Doggystyle_v1.0.safetensors | Weights | 155.1 MB | 422281a53f73 |
| loras/minimaxh3/H3_Motion_BoosterV2.safetensors | Weights | 155.1 MB | f6a6897162b9 |
| loras/minimaxh3/H3_传教士_偏体位.safetensors | Weights | 310.2 MB | 95409b988466 |
| loras/minimaxh3/H3_传教士狗狗视角_偏运动幅度.safetensors | Weights | 310.2 MB | 8451813fa100 |
| loras/minimaxh3/HMNSFW-AIO-V2.5.safetensors | Weights | 86.0 MB | a07732a84fd7 |
| loras/minimaxh3/MysticXXX_MMH3-V1.safetensors | Weights | 596.5 MB | 7dd837f4c542 |
| loras/minimaxh3/bounceV07_fl2va-000230_Intense.safetensors | Weights | 596.4 MB | ed0380ae4c7a |
| loras/minimaxh3/h3_Better_NSFW_Motion_V1.safetensors | Weights | 298.3 MB | 15615bf5aef7 |
| loras/minimaxh3/minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_bf16.safetensors | Weights | 2.0 GB | c396a9a06f58 |
| loras/minimaxh3/moawxx_%E5%A5%B3%E6%80%A7%E5%BF%AB%E6%84%9F%E5%91%BB%E5%90%9F%2B%E8%BA%AB%E4%BD%93%E6%89%AD%E5%8A%A8.safetensors | Weights | 155.1 MB | bc0841e21619 |
| text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors | Weights | 15.7 GB | 35a88d510442 |
| unet/h3ErosMax_beta5_fp8.safetensors | Weights | 14.0 GB | a8067999c655 |
| vae/minimax_h3_audio_vae_fp32.safetensors | Weights | 605.3 MB | 8e505d95dd15 |
| vae/minimax_h3_video_vae_fp16.safetensors | Weights | 5.2 GB | 7c1f131492e7 |
| models-manifest.json | Configuration | 2.4 KB | — |
| README.md | Documentation | 1.5 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 40.1 GB
Released by Xinyu through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 40.1 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.