Back to Model
M

MOSS-Image ComfyUI Components

Model Oct 19, 2026

Comfy-Org's component packaging of MOSS-Image Base and SFT, with BF16 and int8 diffusion weights, Qwen3.5 text encoders and FLUX.2 VAE files.

Visit primary link 2 sources attached

What happened lately

MOSS-Image ComfyUI Components timeline

Comfy-Org repackages MOSS-Image into the folders used by ComfyUI. The repository supplies Base and SFT diffusion models in BF16 and int8 ConvRot variants, Qwen3.5 9B text encoders in BF16, int8 ConvRot and W4A8 GPTQ forms, and two FLUX.2 VAE files. Its English card declares Apache-2.0 licensing.

This is packaging of existing OpenMOSS models, rather than a new foundation model. The repository began October 16, 2026. Its history records a small VAE upload on October 17 and a BF16 text-encoder upload on October 18. The card documents file placement but does not establish a minimum ComfyUI version, complete workflow graph, hardware requirement or measured speed advantage. A file’s quantization label also does not describe the precision or total memory use of the complete pipeline.