MiniMax H3 opens its weights as a multimodal video model with native stereo audio
mark_k · x · 2026-08-04
MiniMax H3 opens its weights as a unified multimodal video model
MiniMax H3 is now open on Hugging Face, and the post describes it as a next-generation open-weights multimodal video model that combines text, images, video, and audio in one context.
Key details:
- Generates video with native stereo audio up to 15 seconds at 2K.
- Strong at instruction following, text rendering, brand rendering, and video-to-video motion transfer.
- Targeted at advertising, branding, e-commerce, product design, and gaming.
- The post claims 2K pricing is less than one-third of mainstream models.
- It uses a 33B-parameter Omni-Transformer conditioned on Qwen3-VL-32B.
- Available checkpoints include FL2VA and Ref2VA, with local inference support via Diffusers, SGLang, vLLM, and native ComfyUI.
The author calls it one of the strongest open video models released so far.
Related event: MiniMax H3 Local Tests: Benchmark Results and Comparisons Across GPUs(88 posts)→
More from Multimodal
- Fan animation made in Blender with Claude Code (Opus 5.5) — Icy-Appointment3583 · 2026-10-16
- Making Blender animations with Opus, Astra and Argon: how you prompt still matters — chongdashu · 2026-10-16
- Tessellation-free geometric queries on displaced surfaces cut caustics variance 10x — ssh4net · 2026-10-16
- Complex multi-link CAD mechanism finally animates, dev celebrates the breakthrough — zetalyrae · 2026-10-16
- OpenMOSS releases MOSS-Image-SFT: Apache 2.0 text-to-image model on Hugging Face — bdsqlsz · 2026-10-16
- Hyperflow open-sourced: 49 sampling steps down to 8, ~3x faster video generation — mhdfaran · 2026-10-16