MOSS-Image Edit HQ | Single and Reference-Guided Editing
Load CLIP
Already loads mossImage_sft_txt_int8.safetensors
ComfyUI model file · Text encoder
MOSS-Image text encoder (Qwen3.5 9B, int8) · Text encoder for MOSS-Image SFT and Base
mossImage_sft_txt_int8.safetensors is the Civitai download name of MOSS-Image’s int8 text encoder, qwen3.5_9b_int8_convrot.safetensors. mossImage_base_txt_int8.safetensors is the same file from the Base version. Put it in ComfyUI/models/text_encoders and load it with Load CLIP, type moss_image.
Comfy-Org’s repo has it as text_encoders/qwen3.5_9b_int8_convrot.safetensors. Either name works; pick it again in Load CLIP if your workflow uses the other.
mossImage_sft_txt_int8.safetensors
The short version
MOSS-Image, from OpenMOSS, conditions its images on Qwen3.5 9B rather than a CLIP or T5 encoder. Comfy-Org packages that encoder in three sizes: bf16 (19.3 GB), int8 convrot (9.75 GB) and w4a8 gptq (7.34 GB). This page is the int8 one.
Civitai’s official MOSS Image listing (model 3005406) carries the same text encoder in both its SFT and Base versions, with the SHA256 32e37874…5a97f025a that matches Comfy-Org’s file. Downloads from that listing were named after the version, mossImage_sft_txt_int8 and mossImage_base_txt_int8, which is why both names are searched. They are one file, so one copy serves both the SFT and Base diffusion models.
Install
Official ComfyUI repackage, no sign-in. Rename if your workflow needs the Civitai name.
saves as qwen3.5_9b_int8_convrot.safetensors
Same file in both the SFT and Base versions. Sign-in may be required.
saves as mossImage_sft_txt_int8.safetensors
ComfyUI/ └─ models/ ├─ … └─ text_encoders/ └─ mossImage_sft_txt_int8.safetensors
$ cd ComfyUI/models/text_encoders$ wget -O mossImage_sft_txt_int8.safetensors \$ https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors$ sha256sum mossImage_sft_txt_int8.safetensors# 32e3787472f708fa5f83624acbf5ef23f7fc89c439e05065ea57cfe5a97f025a
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.
Use
Load CLIP reads the text encoder and outputs a CLIP. Every prompt node in the graph encodes through it.
Set the type
In Load CLIP choose type moss_image. Other types load the weights but encode prompts wrongly.
Less VRAM
qwen3.5_9b_w4a8_gptq.safetensors (7.34 GB) is the smaller MOSS text encoder.
Works with
mossImage_sft_txt_int8.safetensors is one part of the stack. A workflow that reports a missing model usually needs these too.
Run it
We match workflows by the model files they load: this exact file first, then other versions of it, then workflows built for the same use, then other models from the same family.
1 workflow
These workflows already load mossImage_sft_txt_int8.safetensors. Open one and run it as is.
Load CLIP
Already loads mossImage_sft_txt_int8.safetensors
7 workflows
These workflows load a different precision or format of the same model. Select mossImage_sft_txt_int8.safetensors in their Load CLIP node to use this version.
Load CLIP
qwen3.5_9b_w4a8_gptq.safetensors
Replace with mossImage_sft_txt_int8.safetensors
Load CLIP
qwen3.5_9b_w4a8_gptq.safetensors
Replace with mossImage_sft_txt_int8.safetensors
Load CLIP
qwen3.5_9b_w4a8_gptq.safetensors
Replace with mossImage_sft_txt_int8.safetensors
Load CLIP
qwen3.5_9b_w4a8_gptq.safetensors
Replace with mossImage_sft_txt_int8.safetensors
Load CLIP
qwen3.5_9b_w4a8_gptq.safetensors
Replace with mossImage_sft_txt_int8.safetensors
Load CLIP
qwen3.5_9b_w4a8_gptq.safetensors
Replace with mossImage_sft_txt_int8.safetensors
Load CLIP
qwen3.5_9b_w4a8_gptq.safetensors
Replace with mossImage_sft_txt_int8.safetensors
Compare
| File | Relation | Size | Best for |
|---|---|---|---|
| mossImage_sft_txt_int8.safetensors | This page · Civitai name (SFT) | 9.75 GB | MOSS-Image SFT and Base |
| mossImage_base_txt_int8.safetensors | Identical bytes · Civitai name (Base) | 9.75 GB | Same file |
| qwen3.5_9b_int8_convrot.safetensors | Identical bytes · Comfy-Org name | 9.75 GB | Same file |
| qwen3.5_9b_w4a8_gptq.safetensors | Same encoder · w4a8 | 7.34 GB | Less VRAM |
| qwen3.5_9b_bf16.safetensors | Same encoder · bf16 | 19.31 GB | Full precision |
Troubleshoot
Why: The workflow asks for this exact filename, but ComfyUI cannot find it in models/text_encoders. Usually the file is missing, sits in a different folder, or was saved under another name such as qwen3.5_9b_int8_convrot.safetensors.
Fix: Move the file into ComfyUI/models/text_encoders/, rename it to match the workflow (or pick it again in the dropdown), then refresh the node list or restart ComfyUI.
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 9,748,121,532 bytes and that its SHA256 matches the value above. Download it again from Hugging Face · Comfy-Org/MOSS-Image if not.
Rather skip the setup?
Run MOSS-Image Edit HQ in your browser on a cloud GPU. Nothing to download.
FAQ
32e3787472f708fa5f83624acbf5ef23f7fc89c439e05065ea57cfe5a97f025a, in both versions and only the download name changes.moss_image.ComfyUI/models/text_encoders/. Then refresh the node list or restart ComfyUI so Load CLIP lists the file.Sources: Comfy-Org/MOSS-Image · OpenMOSS-Team/MOSS-Image-SFT · Civitai model 3005406
RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.