Comfy-Org Makes Boogu-Image's 10B Open-Source Image Model Plug-and-Play
Comfy-Org repackages the new Boogu-Image-0.1 model family into drop-in ComfyUI-ready files, making the 10B Apache-2.0 image model instantly accessible to node-graph workflows
Takeaways
- Comfy-Org repackaged the Boogu-Image-0.1 model files into ComfyUI-ready
.safetensorsformat on Hugging Face, requiring no conversion. - Boogu-Image-0.1 is a 10B Apache-2.0 image generation and editing model with Base, Turbo, and Edit variants.
- The Turbo variant generates in 3-4 steps using distillation; the fp8 backbone fits in ~10 GB VRAM.
- Architecture uses a double-stream MMDiT with Qwen3-VL-8B as text encoder and the FLUX.1 VAE.
- Key strength is bilingual (Chinese/English) text rendering in posters, stamps, and UI mockups.
- Known gaps include image-to-image consistency, world knowledge, and small-detail artifacts from the FLUX VAE.
Comfy-Org has published a repackaged distribution of the Boogu-Image-0.1 model files on Hugging Face, formatted specifically for ComfyUI. The original Boogu-Image-0.1 family, released by the Boogu project, is a 10B-parameter Apache-2.0 image generation and editing model. Comfy-Org's packaging means ComfyUI users can drop the files directly into their model folders and start using the model without any conversion or custom tooling.
What Comfy-Org actually ships
Comfy-Org ships the diffusion backbone in three formats: bf16 (20.59 GB), fp8 scaled (10.31 GB), and nvfp4 (5.83 GB), plus the text encoder qwen3vl_8b_fp8_scaled (10.59 GB) and the VAE flux1_vae_bf16 (0.17 GB). There is also a distilled Turbo LoRA (1.35 GB) you can apply to Base.
The folder layout is explicit. Place files as follows:
ComfyUI/
└── models/
├── diffusion_models/
│ ├── boogu_image_base_bf16.safetensors
│ ├── boogu_image_base_fp8_scaled.safetensors
│ ├── boogu_image_edit_bf16.safetensors
│ ├── boogu_image_turbo_bf16.safetensors
│ └── boogu_image_turbo_fp8_scaled.safetensors
├── text_encoders/
│ └── qwen3vl_8b_fp8_scaled.safetensors
└── vae/
└── flux1_vae_bf16.safetensors
ComfyUI-Boogu support was released shortly after the base model, with the Boogu team explicitly thanking ComfyUI for the fast integration. ComfyUI core picked up native support for Boogu-Image via commit Support Boogu-Image (CORE-308), making it straightforward to load without any custom nodes.