# minimax_h3_fl2va_pruned_int8_convrot.safetensors

> MiniMax H3 FL2VA pruned (int8 convrot) · Text, image and first-last-frame to video for MiniMax H3

This is MiniMax H3’s main FL2VA model, pruned and quantized to int8 convrot. It handles text to video, image to video and first-last-frame to video, with sound. Put it in ComfyUI/models/diffusion_models and load it with Load Diffusion Model.

Comfy-Org recommends int8_convrot when your PyTorch build has CUDA 13 (cu130). If not, use the fp8_scaled file instead.

## Facts

- **Type:** Diffusion model (video + audio)
- **Format:** Safetensors · int8 convrot
- **Size:** 20.97 GB (20,970,379,616 bytes)
- **SHA256:** e889202c41dafb67b10d67b97f0d8541508036a6090af23425a5c2615d03c47a
- **ComfyUI folder:** ComfyUI/models/diffusion_models/
- **Loader node:** Load Diffusion Model
- **License:** MiniMax H3 Community License

## What is minimax_h3_fl2va_pruned_int8_convrot.safetensors?

FL2VA stands for first-last frame to video and audio. It is the general-purpose MiniMax H3 model: give it only a prompt for text to video, a start image for image to video, or both a first and a last frame to control where the clip begins and ends.

This copy is pruned (unused inference weights removed) and quantized with int8 convrot, a rotation-based int8 format that keeps quality close to bf16 at half the size. It is the file most of our MiniMax H3 workflows load, often with a Turbo LoRA for 4 to 8 steps.

## Download and install

- [Hugging Face · Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_fl2va_pruned_int8_convrot.safetensors) · saves as `minimax_h3_fl2va_pruned_int8_convrot.safetensors` · ComfyUI repackage of the official weights. Recommended.

```sh
cd ComfyUI/models/diffusion_models
wget -O minimax_h3_fl2va_pruned_int8_convrot.safetensors \
  https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_fl2va_pruned_int8_convrot.safetensors
sha256sum minimax_h3_fl2va_pruned_int8_convrot.safetensors
# e889202c41dafb67b10d67b97f0d8541508036a6090af23425a5c2615d03c47a
```

Safetensors files hold tensors only and cannot run code. Still compare the SHA256: a failed download often leaves an HTML error page saved under the model’s filename.

## How to load it in ComfyUI

Load Diffusion Model reads the diffusion model and outputs a MODEL for the sampler. The text encoder and VAE load in their own nodes.


## Files a MiniMax H3 video workflow needs

Use the Ref2VA diffusion model instead of FL2VA for reference-to-video workflows.

- [`qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors`](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors) · Text encoder (Qwen3-VL 32B) · `models/text_encoders/` · 15.69 GB
- [`minimax_h3_video_vae_fp16.safetensors`](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_fp16.safetensors) · Video VAE · `models/vae/` · 5.21 GB
- [`minimax_h3_audio_vae_fp32.safetensors`](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors) · Audio VAE · `models/vae/` · 605 MB
- [`minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_bf16.safetensors`](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/loras/minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_bf16.safetensors) · Turbo LoRA, 4 steps · `models/loras/` · 1.96 GB

## Workflows: Uses this exact file

These workflows already load minimax_h3_fl2va_pruned_int8_convrot.safetensors. Open one and run it as is.

- [Pika H3 Flex ComfyUI Workflow | 8-Step Ref2Video](https://www.runcomfy.com/comfyui-workflows/pika-h3-flex-comfyui-8-step-ref2video-stereo-audio)
- [MiniMax H3 Veda Acceleration | 3x Fast Ref-to-Video](https://www.runcomfy.com/comfyui-workflows/minimax-h3-veda-acceleration)
- [MiniMax H3 Camera Control ComfyUI | Orbit, Dolly, Elevation](https://www.runcomfy.com/comfyui-workflows/minimax-h3-camera-control-comfyui-orbit-dolly-elevation)
- [MiniMax H3 Singularity Reference To Video ComfyUI Workflow](https://www.runcomfy.com/comfyui-workflows/minimax-h3-singularity-reference-to-video-comfyui)
- [VDN H3 ComfyUI Text To Video |  Faster MiniMax H3 VDN 8-Step Workflow wihout LoRA](https://www.runcomfy.com/comfyui-workflows/vdn-h3-comfyui-text-to-video-8-step-video-and-audio)
- [MiniMax FastH3 4-Step ComfyUI Workflow | Super Fast Minimax H3 Workflow](https://www.runcomfy.com/comfyui-workflows/minimax-fasth3-4-step-comfyui-workflow-minimax-h3-t2v-and-i2v)
- [MiniMax H3 Fun Control Workflow | Controlled Image-to-Video](https://www.runcomfy.com/comfyui-workflows/minimax-h3-fun-control-in-comfyui-depth-and-pose-video)
- [MiniMax H3 First Last Frame to Video Acceleration](https://www.runcomfy.com/comfyui-workflows/minimax-h3-flf2v-in-comfyui-turbo-lora-stereo-audio)
- [MiniMax H3 ComfyUI Image to Video 4-step Turbo](https://www.runcomfy.com/comfyui-workflows/minimax-h3-comfyui-i2v-4-step-turbo-stereo-audio)
- [MiniMax H3 ComfyUI Text-to-Video 4-step Turbo](https://www.runcomfy.com/comfyui-workflows/minimax-h3-comfyui-text-to-video-4-step-turbo)
- [MiniMax H3 Image To Video ComfyUI - Cinematic](https://www.runcomfy.com/comfyui-workflows/minimax-h3-image-to-video-comfyui-native-audio)
- [MiniMax H3 T2V ComfyUI - AI Text-to-Video Generator](https://www.runcomfy.com/comfyui-workflows/minimax-h3-t2v-comfyui-video-and-native-stereo-audio)

## Workflows: Built for first-last-frame video

These workflows load another first-last-frame video model in the same folder, or they are first-last-frame video workflows that already have a Load Diffusion Model node. Select minimax_h3_fl2va_pruned_int8_convrot.safetensors there.

- [MiniMax H3 8-Step Acceleration | All-in-One Video Workflow](https://www.runcomfy.com/comfyui-workflows/minimax-h3-8-step-acceleration-all-in-one-comfyui-workflow)
- [Minimax H3 I2V Dual Clock 8-Step| Ultra-Fast High-Quality Video](https://www.runcomfy.com/comfyui-workflows/minimax-h3-i2v-comfyui-8-step-dual-clock-audio-video)

## Same file, other names, and alternatives

- `minimax_h3_fl2va_pruned_int8_convrot.safetensors` · This page · pruned int8 convrot · 20.97 GB · Default pick on PyTorch cu130
- `minimax_h3_fl2va_pruned_fp8_scaled.safetensors` · Same model · fp8 scaled · 20.96 GB · When int8_convrot is not supported
- `minimax_h3_fl2va_pruned_w6a8.safetensors` · Same model · w6a8 · 15.98 GB · Less VRAM
- `minimax_h3_fl2va_pruned_bf16.safetensors` · Same model · bf16 · 40.23 GB · Highest fidelity
- `minimax_h3_ref2va_pruned_fp8_scaled.safetensors` · Sibling model · Ref2VA · 20.96 GB · Reference-guided video

## Common errors and fixes

### Value not in list: unet_name: 'minimax_h3_fl2va_pruned_int8_convrot.safetensors' not in [...]

Why: The workflow asks for this exact filename, but ComfyUI cannot find it in models/diffusion_models. Usually the file is missing, sits in a different folder.

Fix: Move the file into ComfyUI/models/diffusion_models/, rename it to match the workflow (or pick it again in the dropdown), then refresh the node list or restart ComfyUI.

### Error loading int8_convrot weights / unsupported quantization format

Why: int8 convrot needs a recent ComfyUI and a PyTorch build with CUDA 13 (cu130).

Fix: Update ComfyUI and PyTorch, or use minimax_h3_fl2va_pruned_fp8_scaled.safetensors instead.

### SafetensorError: Error while deserializing header

Why: The download is incomplete, or the browser saved an HTML sign-in or error page under the model’s filename.

Fix: Check that the file is exactly 20,970,379,616 bytes and that its SHA256 matches the value above. Download it again from Hugging Face · Comfy-Org/MiniMax-H3 if not.

## FAQ

### What does “pruned int8 convrot” mean?

Pruned: weights not used at inference are removed. int8 convrot: weights are stored in 8 bits after a rotation that keeps outliers in check, so quality stays close to bf16 at half the size.

### Where do I put minimax_h3_fl2va_pruned_int8_convrot.safetensors in ComfyUI?

In `ComfyUI/models/diffusion_models/`. Then refresh the node list or restart ComfyUI so Load Diffusion Model lists the file.

### Which ComfyUI node loads minimax_h3_fl2va_pruned_int8_convrot.safetensors?

[Load Diffusion Model](/comfyui-nodes/ComfyUI/unet-loader). It is built into ComfyUI, so no custom nodes are needed.

### Can I use minimax_h3_fl2va_pruned_int8_convrot.safetensors commercially?

Yes, within the MiniMax H3 Community License. Commercial products must show “MiniMax H3” in their interface, and companies with more than US$20 million in yearly revenue need written permission from MiniMax first. The license also limits use to its listed territories, so read it before shipping.

## Sources

- [Comfy-Org/MiniMax-H3 on Hugging Face](https://huggingface.co/Comfy-Org/MiniMax-H3)
- [MiniMaxAI/MiniMax-H3 (original weights)](https://huggingface.co/MiniMaxAI/MiniMax-H3)
- [ComfyUI MiniMax H3 tutorial](https://docs.comfy.org/tutorials/video/minimax/minimax-h3)

Canonical page: https://www.runcomfy.com/comfyui-workflows/model-files/minimax_h3_fl2va_pruned_int8_convrot.safetensors
