AI News: Things Will Never Be The Same
An image model generating 4K images in about 10 seconds, available via API, MiniMax code and Hugging Face, with a Hugging Face space. The creator browsed sample images and said it looks good.
Open-weights MiniMax image generation and editing model producing up to ~16 MP images in 6 sampling steps.
Updated EU status checked Last shown
MiniMax H3 Image at a glance
- What is MiniMax H3 Image?
- MiniMax H3 Image is a model from MiniMax. Open-weights MiniMax image generation and editing model producing up to ~16 MP images in 6 sampling steps.
- Who makes MiniMax H3 Image?
- MiniMax H3 Image is made by MiniMax.
- Does MiniMax H3 Image work in the EU?
- Available in the EU. Weights are downloadable from Hugging Face and the global API and Hailuo AI web app are offered; no EU exclusion was found in the provided material, though the custom community license text was not reviewed. Last verified 14 Nov 2026.
- How much does MiniMax H3 Image cost?
- Open source. Weights free to download under community license; also available via paid MiniMax API and Hailuo AI app (pricing not given in provided material). Licence: MiniMax-H3-Image Community License Agreement (custom).
- Where was MiniMax H3 Image shown?
- In 3 creator videos by Matt Wolfe, AI Search, Matthew Berman. Most recently by Matt Wolfe on 13 Nov 2026, at 24:05 in “AI News: Things Will Never Be The Same”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
About
MiniMax-H3-Image is an image generation and editing model from the MiniMax H3 family. A single checkpoint handles text-to-image and editing of one or more reference images. It uses a diffusion transformer with the H3 architecture minus the audio layers, and the MiniMax-H3 text encoder unchanged. The checkpoint is distilled to 6 sampling steps with no classifier-free guidance.
It generates images of about 16 megapixels at various aspect ratios, from 4096x4096 (1:1) to 6272x2688 (21:9), without an upscaler. Creators report 4K output in under 10 seconds, support for up to 16 reference images, and GGUF/quantized versions (around 18 GB, versus roughly 68 GB for the full weights) that run on as little as 6 GB VRAM. In comparisons with Qwen Image 2.1 and Krea 2, reviewers found it strong at editing, UI design and magazine covers, but weaker on some anatomy and realism.
It can be used via the MiniMax API (text-to-image, image editing and prompt-expansion endpoints), the Hailuo AI web app, the MiniMax desktop agent, or run locally from Hugging Face weights, including through ComfyUI templates. The weights are under a custom MiniMax-H3-Image community license, not a standard open-source license.
Key features
- Unified text-to-image and multi-reference image editing in one checkpoint
- About 16 MP output (up to 4096x4096 or 6272x2688) with no upscaler
- Distilled to 6 sampling steps, no classifier-free guidance
- Reported support for up to 16 reference images
- Open weights with quantized/GGUF variants for low-VRAM local use
- ComfyUI workflow templates and MiniMax API access
3 moments
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
New top local AI image editor is here!
Open image generation and editing model that can make a 4K image in under 10 seconds, supports up to 16 reference images and runs with as little as 6 GB VRAM. Compared against Qwen Image 2.1 and Krea 2, it is strong at editing, UI design and magazine covers but weaker on some anatomy and realism. Creator walks through ComfyUI install, text-to-image and edit workflows, and GGUF versions for low VRAM.
MA
Gemini Carbon, NVIDIA CEO AFFAIR, Fable 5.2, Meta Muse Ember, GPT Image 2.6, Minimax H3: AI NEWSMatthew Berman11 Nov 2026
Fully open-weights image generation model. It is 68 GB, with an 18 GB quantized version that most computers can run.
Keep exploring
Similar AI tools
Closest neighbours by meaning, not just by category.
MH
MiniMax H3
Model · MiniMax
Open-weight video generation model from MiniMax, part of its H3 family alongside H3 Image.
UnverifiedVideo Generation
Release date n/a
PI
P-H3 Image
Model · Pruna AI
Open-weight image generation and editing model from Pruna AI, distilled from MiniMax's H3 video model to run in 8 steps.
Available in EUImage Generation
Release date n/aOpen source
HM
H3 Max
API · fal
Fine-tuned, low-latency variant of MiniMax H3 video model served through fal's paid API.
UnverifiedVideo Generation
Release date n/aAPI
FA
FastH3
Model · Hao AI Lab
Open-weight 4-step sparse-distilled version of MiniMax H3 for fast text-to-video-and-audio generation on Blackwell GPUs.
Available in EUVideo Generation
Release date n/aOpen source
MH
MiniMax H3 image workflow
App
Community ComfyUI workflows that use the MiniMax H3 video model with a one-frame count to output still images.
UnverifiedImage Generation
4 Aug 2026Open source
MH
MiniMax H3 audio generation
Research · MiniMax
Open-weight multimodal model from MiniMax whose video pipeline can be run audio-only to generate music and sound effects.
UnverifiedVoice & Audio
Release date n/a