Unverified50% confidenceReleaseExact time
ComfyUI-INT4-Fast是将INT4（W4A4）量化推理带入ComfyUI生态的自定义节点包，使6GB显存的RTX 3060也能运行基于Flux的Krea2 Turbo模型
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/12/2026
First Seen
Valid until: 10/10/2026
Sources
ComfyUI-INT4-Fast：6GB显卡流畅运行Flux大模型实测
redditr/StableDiffusion7/10/2026
Related Claims
UnverifiedComfyUI-INT4-Fast在BobJohnson24的ComfyUI-INT8-Fast基础上独立改进而来，底层依托comfy-kitchen调用GPU的Tensor Core69% similarUnverifiedComfyUI-INT4-Fast支持原生混合精度检查点，按逐层粒度决定计算路径：主干模块路由至INT4 Tensor Core，敏感层（patch projection）以INT8格式导向Triton的INT8执行路径64% similarUnverified有开发者将多个主流图像模型转换为INT4 ConvRot W4A4格式并适配最新版本的ComfyUI63% similarUnverifiedTFT_eSPI是由Bodmer开发的高性能Arduino图形库，针对ESP8266和ESP32平台深度优化，支持超过40种TFT驱动芯片并内置Sprite功能63% similarUnverifiedRTX 6000 Ada的第四代Tensor Core支持FP8精度运算62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/489319API
curl https://kongchang.com/api/v1/knowledge/claims/489319MCP
get_claim(id=489319)