PrunaAI releases the P-H3-Image image model, fine-tuned from MiniMax H3
PrunaAI releases the P-H3-Image image model, fine-tuned from MiniMax H3, supporting 8 MP text-to-image generation and editing with up to 8 input images. The model is a 33.1B-parameter transformer, uses 8-step sampling without CFG, includes a built-in Qwen3.5-9B prompt upsampler, and is natively supported in ComfyUI v0.44.0.