Hugging Face's logo

Comfy-Org
/
MiniMax-H3

Diffusion Single File
comfyui
Model card Files
xet
 Community
67
MiniMax-H3
File size: 2,500 Bytes
3f57e82
cfc0a7e
 
 
3f57e82
 
eb8a161
dcc9da7
 
3f57e82
 
 
 
 
 
44aa1aa
 
cec22ac
 
3f57e82
d07f69b
93acf8c
d07f69b
014cd40
d07f69b
014cd40
d07f69b
014cd40
 
 
3f57e82
 
 
 
 
 
 
 
0bd506d
3f57e82
0543966
3f57e82
 
0bd506d
0543966
 
3f57e82
 
 
 
cec22ac
 
 
 
3f57e82
 
 
0e6104a
 
 
 
fd70b39
 
8738f4e
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
---
license: other
license_name: minimax-h3-community-license-agreement
license_link: https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE
tags:
- comfyui
- diffusion-single-file
base_model:
- MiniMaxAI/MiniMax-H3
---

# MiniMax H3

Repackaged model files for ComfyUI.

Original model repository:

- https://huggingface.co/MiniMaxAI/MiniMax-H3
- https://huggingface.co/lightx2v/Minimax-h3-Turbo

The Qwen3-VL-32B `nvfp4_awq` quant is converted from: https://huggingface.co/cybermotaz/Qwen3-VL-32B-Instruct-NVFP4

This `nvfp4` text encoder does not require Blackwell GPU to use.

For diffusion models prefer int8_convrot if you are able to use pytorch with cu130.

`fp8_scaled` should only be used if you cannot use `int8_convrot`.

---

Place the files in the following folders:

```
πŸ“‚ ComfyUI/
β”œβ”€β”€ πŸ“‚ models/
β”‚   β”œβ”€β”€ πŸ“‚ diffusion_models/
β”‚   β”‚   β”œβ”€β”€ minimax_h3_fl2va_bf16.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_fl2va_int8_convrot.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_fl2va_pruned_bf16.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_fl2va_pruned_int8_convrot.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_fl2va_pruned_fp8_scaled.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_ref2va_bf16.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_ref2va_int8_convrot.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_ref2va_pruned_bf16.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_ref2va_pruned_int8_convrot.safetensors
β”‚   β”‚   └── minimax_h3_ref2va_pruned_fp8_scaled.safetensors
β”‚   β”œβ”€β”€ πŸ“‚ text_encoders/
β”‚   β”‚   β”œβ”€β”€ qwen3vl_32b_minimax_h3_bf16.safetensors
β”‚   β”‚   β”œβ”€β”€ qwen3vl_32b_minimax_h3_int8_convrot.safetensors
β”‚   β”‚   └── qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors
β”‚   β”œβ”€β”€ πŸ“‚ loras/
β”‚   β”‚   β”œβ”€β”€ minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_bf16.safetensors
β”‚   β”‚   β”œβ”€β”€ minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors
β”‚   β”‚   └── minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors
β”‚   β”œβ”€β”€ πŸ“‚ vae/
β”‚   β”‚   β”œβ”€β”€ minimax_h3_audio_vae_fp32.safetensors
β”‚   β”‚   └── minimax_h3_video_vae_fp16.safetensors
```

## Workflows

- I2V：https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_i2v.json
- T2V：https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_t2v.json
- R2V：https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_r2v.json