---
title: WanVideo_comfy_fp8_scaled
canonical_url: "https://www.modelscope.cn/models/Kijai/WanVideo_comfy_fp8_scaled"
md_url: "https://www.modelscope.cn/models/Kijai/WanVideo_comfy_fp8_scaled.md"
repository: Kijai/WanVideo_comfy_fp8_scaled
last_updated: 2026-06-14
license: apache-2.0
base_model:
  - Wan-AI/Wan2.1-VACE-14B
  - Wan-AI/Wan2.1-VACE-1.3B
base_model_relation: finetune
parameters: 924.6B
tensor_type:
  - F32
  - U8
  - F8_E5M2
  - F8_E4M3
  - BF16
  - F16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 68481
stars: 68
tags:
  - diffusion-single-file
  - comfyui
---

# WanVideo_comfy_fp8_scaled

> WanVideo_comfy_fp8_scaled - Kijai 在 ModelScope 开源的模型。Better fp8 scaled models (when measured against fp16) based on quantization code from https://github.com/Tencent-Hunyuan/HunyuanVideo/blob/main/hyvideo/modules/fp8optimization.py

Kijai/WanVideo_comfy_fp8_scaled 是 ModelScope 魔搭社区上的 924.6B 参数机器学习模型，采用 apache-2.0 许可，基于 Wan-AI/Wan2.1-VACE-14B、Wan-AI/Wan2.1-VACE-1.3B 构建。

- **Repository**: Kijai/WanVideo_comfy_fp8_scaled
- **License**: apache-2.0
- **Parameters**: 924.6B
- **Base model**: Wan-AI/Wan2.1-VACE-14B, Wan-AI/Wan2.1-VACE-1.3B
- **Tags**: diffusion-single-file, comfyui
- **Downloads**: 68481
- **Stars**: 68
- **Last updated**: 2026-06-14

Source: https://www.modelscope.cn/models/Kijai/WanVideo_comfy_fp8_scaled

---

Better fp8 scaled models (when measured against fp16) based on quantization code from https://github.com/Tencent-Hunyuan/HunyuanVideo/blob/main/hyvideo/modules/fp8_optimization.py

Can be used with: https://github.com/kijai/ComfyUI-WanVideoWrapper (latest version) and ComfyUI native WanVideo nodes.

14B-T2V comparison test without LoRAs, 25 steps, 832x480x81
---

<video controls autoplay width=50% src=https://cdn-uploads.huggingface.co/production/uploads/63297908f0b2fc94904a65b8/DwlAGbj20it1unZW54NDC.mp4></video>

2.2 A14B-T2V test
---

<video controls autoplay width=50% src=https://cdn-uploads.huggingface.co/production/uploads/63297908f0b2fc94904a65b8/6A_AZ7GN_uxeRH0vwsWkH.mp4></video>

<video controls autoplay width=50% src=https://cdn-uploads.huggingface.co/production/uploads/63297908f0b2fc94904a65b8/GpuqQ4YwoR3kjxkhuvP8P.mp4></video>

The e5m2 marked as v2 is the one uploaded here and these are all scaled even if I forgot to label properly.