MiniMax releases open-weight H3-Image model generating 4K images in under 10 seconds

MiniMax released H3-Image, a 33.1 billion parameter open-weight model that unifies text-to-image generation and editing, on November 11, 2026, according to the MiniMax blog post.

Technical details

• The diffusion transformer has 33.1 billion parameters and is built on MiniMax's H3 architecture
• It generates a 4K (4096x4096, about 16 megapixel) image in under 10 seconds
• Supported resolutions range from 4096x4096 up to 6272x2688
• The model is distilled to 6 sampling steps with no classifier-free guidance
• It supports up to 16 reference images for editing and consistency tasks
• Text-to-image generation and image editing are unified in a single checkpoint

What it means for people who build with AI

The release gives creators and developers an open-weight alternative for fast, high-resolution image generation and editing that can be self-hosted rather than accessed only through closed APIs. MiniMax says its Highspeed mode runs 3 to 4 times faster than mainstream alternatives, and up to 5 times faster overall for 4K output, which matters for workflows that need quick iteration at high resolution.

Availability and licensing

H3-Image is available through the Hugging Face model page, the MiniMax API platform, MiniMax Code, and Hailuo AI, with a launch-week 50% discount for 4K-plus resolution. It is released under the MiniMax-H3-Image Community License Agreement, and commercial use requires separate written authorization.

Sources

• MiniMax H3-Image: Bringing 4K Image Generation into the 10-Second Era (MiniMax, November 11, 2026, official)
• MiniMaxAI/MiniMax-H3-Image (Hugging Face)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.