Reactor Exclusive Preview: Open-Source Audio-Video Model FastH3-Omni

On October 14, 2026, Reactor introduced FastH3-Omni, an open-source model built on MiniMax's realtime H3 model and post-trained by FastVideo (haoailab) with NuvA Lab. Reactor describes it as the fastest and highest-quality open-source version of realtime H3 currently available.

The model supports first and last frame conditioning and audio input, allowing users to generate a video with synchronized audio from a single prompt. It is available for exclusive preview on the Reactor API until October 20, 2026.