MiniMaxLaunchAI ModelsJuly 29, 2026

MiniMax H3 omni-modal model launches with open weights

MiniMax H3 is a general-purpose, omni-modal generative system supporting text, image, video, and audio, generating video with native stereo audio up to 2K and 15 seconds. Open weights are available on Hugging Face, with integrations on ComfyUI, Picsart, LeonardoAi, and Vercel AI Gateway.

How this story unfolded

3 weeks · 8 reports · 32 community posts · 40 of 46 shown

Jul 31
- MiniMax H3 now available on Vercel AI Gatewayvercel.com
Aug 1
- Open source video (H3) model: Finally some water in desert.
Aug 2
Aug 3
Aug 5
Aug 6
- Kijai/MiniMax-H3-TAEhuggingface.co
- Kijai/MiniMax-H3-experimentalhuggingface.co
Aug 7
- Kijai/MiniMax-H3_comfyhuggingface.co
Aug 8
- For 8GB (& below) Vram Rockers, kijay released a 4 bit version of minimax
Aug 10
- One of the best programmers on Earth just wrote a MiniMax H3 inference engine for Mac computers.🫨...
Aug 13
- Minimax H3 ♥
Aug 16
- MiniMax H3 deserves way more attention. It's the first video model that makes the sound and the...
Aug 19
- LBH-123-AI/Minimax_h3_latent_Upscalerhuggingface.co
Aug 20
- Teaming up with @ComfyUI for the H3 challenge🏄‍♀️ Build, submit by 9/1, win an RTX 5090 or 5060...
Aug 22
- MiniMax-H3 Pruned Ref-Delta Fused r1024 — INT8 and INT8 ConvRot ComfyUI versions

MiniMax by email

Get an email when MiniMax has news

No news that day, no email.

MiniMax H3 omni-modal model launches with open weights

How this story unfolded

MiniMax by email

More stories today