Skip to main content
← Back to news

P-Image-OSS: 8-megapixel images in 8 steps, 83 to 101 GB of VRAM to run them locally

Model releases
PrunaAI releases P-Image-OSS, an open image generation and editing model based on MiniMax-H3: 8 steps without CFG, 8 MP text-to-image, editing with up to 8 input images and an included prompt upsampler (Qwen3.5-9B). In bf16 it needs ~101 GB of VRAM (83 without the upsampler); the int8/4-bit ComfyUI variant runs in 21 GB.

Get the next local-AI briefing

Once a week, the top selection from iaLocale: articles, news and resources on local AI, and more.

After signing up you receive a confirmation email (double opt-in). Without confirmation we send nothing. No spam, we promise.