FastVideo Releases Open Source Version of MiniMax Realtime H3 on October 20

AI8 days ago

A collaborative effort between FastVideo, Nuva Lab, and Reactor has produced FastH3-Omni, a post-trained version of the H3 architecture developed by MiniMax AI. The team describes the tool as the fastest and highest-quality open-source implementation of realtime H3, featuring first and last frame conditioning and audio input for strong dialogue. The model supports up to 3 reference images to achieve what the developers characterize as frontier-level performance.

The project is currently available for a limited preview via the Reactor API. FastVideo will release the final model as fully open source on October 20, providing the associated training code and official support for RTX GPUs and DGX spark hardware.

Mapped to your portfolio

Possible scenarios

45%Open Adoption Accelerates+0.5%
25%Commercial Impact Limited-0.1%
18%Policy Scrutiny Rises-0.3%

Each figure is the modeled move of the S&P 500 if the outcome occurs. For information only. Not investment advice.

Looking for more context or analysis?

Tilt chat provides reliable, tailored insights, alerts, and automations.

FastVideo Releases Open Source Version of MiniMax Realtime H3 on October 20

In this story

FastVideo MiniMax Nuva Lab Reactor

Portfolio

SPY
State Street SPDR S&P 500 ETF Trust
100%

Portfolio

SPY
State Street SPDR S&P 500 ETF Trust
100%

Mapped to your portfolio

Possible scenarios

45%Open Adoption Accelerates+0.5%
25%Commercial Impact Limited-0.1%
18%Policy Scrutiny Rises-0.3%

Each figure is the modeled move of the S&P 500 if the outcome occurs. For information only. Not investment advice.