FastH3-Omni FastH3-Omni is the fastest and highest-quality open-source version of MiniMax's realtime H3 world model, post-trained for generating video with synchronized native audio from a single prompt. It supports first-and-last-frame conditioning, audio input with strong dialogue, up to three reference images, and frontier-level performance. The model is available as an exclusive preview on the Reactor API.

Visit product

Featured media

Cosigns Received

Shokunin | Product & R&D
“I’ve been playing with the platform and I really like it product. Will build a product on it soon”
Source
Will Lin
“It's amazing to work with you guys and collab on this!”
Source
Topper
“Cool stuff. I ran a comparison against my own GPUs for a some clips I ran recently: https://toppingdesign.com/writing/what-looks-back-h3-vs-omni/”
Source
MiniMax
“FastH3-Omni on Reactor - video with synchronized audio, from a single prompt. Your daydreams have a sound department now.”
Source
Hao AI Lab
“FastVideo team excited to announce FastH3-Omni! 🚀🚀 Postrained in collaboration with Nuva Lab @nuvalab and Reactor @reactorworld! The final FastH3-Omni model will be fully open sourced on October 20, along with training code and full support on DGX spark, and RTX GPUs! Please checkout @reactorworld's API access, available today!”
Source
FastH3-Omni | Cosign