--- base_model: OpenMOSS-Team/MOSS-Image-SFT base_model_relation: quantized license: apache-2.0 pipeline_tag: text-to-image library_name: gguf tags: - gguf - moonbox - moss-image - text-to-image - diffusion-single-file --- # MOSS-Image (SFT) GGUF GGUF builds of [MOSS-Image (SFT)](https://huggingface.co/OpenMOSS-Team/MOSS-Image-SFT), made by Moonbox for on-device use. Each precision is one file at the root of this repository; pick one by the memory you have. Only the model's own weights are converted: the other parts of the pipeline (text encoder, VAE, tokenizer) come from the original repository unless the section below says otherwise. ## Files The last column is how far each file is from the original BF16 weights: the relative RMS error of its dequantized weights, averaged over all parameters. Image-quality scores against BF16 outputs are added as each precision passes the image eval. | Precision | File | Size | sha256 | Checked against BF16 | |---|---|---|---|---| | F16 | [`moss-image-F16.gguf`](https://huggingface.co/moonbox-ai/MOSS-Image-SFT-GGUF/blob/main/moss-image-F16.gguf) | 10.17 GB | `f1c3f1d9fb5d` | 0.0000005 % (5.3e-9) | | BF16 | [`moss-image-BF16.gguf`](https://huggingface.co/moonbox-ai/MOSS-Image-SFT-GGUF/blob/main/moss-image-BF16.gguf) | 10.17 GB | `3b536d5ed977` | exact | | Q8_0 | [`moss-image-Q8_0.gguf`](https://huggingface.co/moonbox-ai/MOSS-Image-SFT-GGUF/blob/main/moss-image-Q8_0.gguf) | 5.69 GB | `850308d0d7ff` | 0.52 % | | Q6_K | [`moss-image-Q6_K.gguf`](https://huggingface.co/moonbox-ai/MOSS-Image-SFT-GGUF/blob/main/moss-image-Q6_K.gguf) | 4.53 GB | `dce216888fdd` | 1.7 % | | Q5_K_M | [`moss-image-Q5_K_M.gguf`](https://huggingface.co/moonbox-ai/MOSS-Image-SFT-GGUF/blob/main/moss-image-Q5_K_M.gguf) | 4.04 GB | `68a927d6a49e` | 3.1 % | | Q4_K_M | [`moss-image-Q4_K_M.gguf`](https://huggingface.co/moonbox-ai/MOSS-Image-SFT-GGUF/blob/main/moss-image-Q4_K_M.gguf) | 3.57 GB | `f7139199aabb` | 5.7 % | ## How to run These files are the MOSS-Image (SFT) diffusion transformer only. Text encoder (Qwen3.5 9B) and VAE are separate: `Qwen3.5-9B-Q8_0.gguf` from unsloth/Qwen3.5-9B-GGUF, `flux2-vae.safetensors` from Comfy-Org/MOSS-Image (vae/) = FLUX.2's VAE. **stable-diffusion.cpp** (the runtime our benchmark uses, commit 465adf3): ``` sd-cli --diffusion-model moss-image-Q8_0.gguf --llm Qwen3.5-9B-Q8_0.gguf --vae flux2-vae.safetensors \ --tokenizer tokenizer.json -p "a lovely cat holding a sign says 'moss-image.cpp'" --cfg-scale 3.0 -W 1024 -H 1024 ``` Pass MOSS-Image's own `tokenizer.json` (from [OpenMOSS-Team/MOSS-Image-SFT](https://huggingface.co/OpenMOSS-Team/MOSS-Image-SFT/tree/main/tokenizer)) with `--tokenizer`: Qwen3.5 uses a different vocabulary from the embedded one. Defaults: 50 steps, CFG 3, flow shift 5. **ComfyUI** with [ComfyUI-GGUF](https://github.com/city96/ComfyUI-GGUF): the stock ComfyUI-GGUF loader does not know `moss_image` yet. **Which file:** Q8_0 is the closest to the original at about half the size; Q6_K and Q5_K_M trade a little fidelity for memory; Q4_K_M is the smallest. BF16 / F16 are the unquantized weights. **License.** Apache-2.0 (in this repository). ## How these were made From [MOSS-Image (SFT)](https://huggingface.co/OpenMOSS-Team/MOSS-Image-SFT) at revision `f972bd4b4b`: the original BF16 weights converted to GGUF F16 / BF16, then quantized with llama.cpp's `llama-quantize`. Every file was checked against the original before upload; the `sha256` column above is what the files must hash to. ## License Apache-2.0. These files are a derivative of the original model and carry its license: it is copied in this repository, and it governs your use of these files. The original model card follows unchanged. --- # Original model card # MOSS-Image-SFT

User guide · 中文用户指南 · ComfyUI

![MOSS-Image-Base and MOSS-Image-SFT](assets/teaser.webp) MOSS-Image-SFT is a supervised fine-tuned text-to-image model in the **MOSS-Image** family from OpenMOSS, derived from [MOSS-Image-Base](https://huggingface.co/OpenMOSS-Team/MOSS-Image-Base). It uses the same architecture and inference pipeline, supporting natural-language prompts, tag sequences, and JSON layouts for controlling text placement and object arrangement. ## Base vs. SFT In this example, SFT produces clearer text grouping and a more organized whiteboard layout. | MOSS-Image-Base | MOSS-Image-SFT | | --- | --- | | ![MOSS-Image-Base](assets/comparison/base.webp) | ![MOSS-Image-SFT](assets/comparison/sft.webp) | ## Quick start Use Python 3.10 or later and an NVIDIA GPU with CUDA support. Install a PyTorch 2.9 or later build compatible with your CUDA environment, then install: ```bash pip install "diffusers>=0.38" "transformers>=5.3" "accelerate>=1.10" "safetensors>=0.6" "Pillow>=10" ``` ```python import torch from diffusers import DiffusionPipeline pipe = DiffusionPipeline.from_pretrained( "OpenMOSS-Team/MOSS-Image-SFT", custom_pipeline="pipeline_moss_image", trust_remote_code=True, torch_dtype={"default": torch.bfloat16, "vae": torch.float32}, ) pipe.scheduler = type(pipe.scheduler).from_config(pipe.scheduler.config) pipe.to("cuda") prompt = '''A wide view of a refrigerator beside a kitchen counter with a fruit bowl. A small sticky note on the refrigerator door is handwritten in Chinese: "记得买牛奶和鸡蛋！牛奶两盒，鸡蛋一打。顺便带一袋面包。"''' image = pipe( prompt, height=1024, width=1024, num_inference_steps=50, guidance_scale=3.0, uncond_skip_layers=[22], generator=torch.Generator("cpu").manual_seed(42), ).images[0] image.save("refrigerator_note.png") ``` The model and custom pipeline code are downloaded and cached automatically. You can also pass a complete local model directory to `from_pretrained`. The transformer and text encoder use bfloat16; the VAE and initial noise use float32. The bundled VAE uses the FLUX.2 small decoder. For the Gradio interface, CPU offload, and compilation options, follow the [project quick start](https://github.com/OpenMOSS/MOSS-Image#quick-start). For ComfyUI, search for **MOSS-Image** in **ComfyUI Manager** to install the native plugin. Integration into the official ComfyUI repository is in progress. See the [installation and usage guide](https://github.com/OpenMOSS/MOSS-Image/blob/main/comfyui/README.md) and [example workflow](https://github.com/OpenMOSS/MOSS-Image/blob/main/comfyui/moss-image/example_workflows/moss-image.json). ## Sampling settings | Setting | Default | | --- | --- | | Sampler | Flow-matching Euler | | Steps | 50 | | CFG | 3 | | Shift | 5 | | Layer skipping | Negative branch, layer index 22 (zero-based) | | Negative prompt | Empty | The default guidance uses two transformer forwards per step. Pass `uncond_skip_layers=[]` to disable negative-branch skipping, or set `negative_prompt` to provide a negative prompt. Python API image dimensions must be positive multiples of 16. See the [user guide](https://github.com/OpenMOSS/MOSS-Image/blob/main/USERGUIDE.md) for resolution, sampling, prompt, and LoRA training recommendations. ## Prompt enhancer skill The [MOSS-Image Prompt Enhancer](https://github.com/OpenMOSS/MOSS-Image/blob/main/skills/moss-image-prompt-enhancer/SKILL.md) turns image ideas into detailed captions or spatial Layout prompts. It is designed for GPT-6-family agents with local file and Python tools. The bundled script converts pixel-coordinate drafts into compact JSON with integer centered coordinates; it does not call an image-generation service. Install the [skill folder](https://github.com/OpenMOSS/MOSS-Image/tree/main/skills/moss-image-prompt-enhancer) in your agent's skills directory, then invoke `$moss-image-prompt-enhancer` with your image idea, desired caption or Layout output, and any canvas dimensions. Paste the compiled result into MOSS-Image's positive prompt; for Layout, use the same generation dimensions. ## Layout-to-image generation Pass a JSON layout as the prompt to describe the scene, rendered text, objects, and their positions. Layouts use `summary.caption` and `items`, with `bbox_format: "yxyx"` and `bbox_coord_space: "abs_centralized"`. With the Python API, serialize the layout using `json.dumps(layout, ensure_ascii=False)` and pass it to `pipe`. In Gradio, paste the JSON into the Prompt field and set the matching image dimensions. See the [layout format and example](https://github.com/OpenMOSS/MOSS-Image/blob/main/USERGUIDE.md#layout-prompts-experimental). Layouts can also support controlled editing and user-friendly layout tools. Edit text, element descriptions, and bounding boxes to adjust image content and composition in a structured way. ![Layout control](assets/showcase/layout-bboxes.webp) ## Showcase ![Examples overview](assets/showcase/overview.webp) ![Examples overview 2](assets/showcase/overview2.webp) ## Citation ```bibtex @misc{ai2026mossimagetechnicalreport, title={MOSS-Image Technical Report}, author={MOSI. AI}, year={2026}, eprint={2610.23374}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2610.23374}, } ``` ## License and acknowledgements The model weights and bundled inference code are released under the [Apache License, Version 2.0](LICENSE). We thank the Diffusers, Qwen, and FLUX teams.