fal releases H3 Max, a post-trained MiniMax H3 video model that generates 5-second clips in about 3 seconds
- fal released H3 Max, a post-trained model built on the open-weights MiniMax H3, for text-to-video and image-to-video generation via its API, Playground, and Agent.
- fal says H3 Max generates a five-second video in about three seconds, claiming roughly 35x the throughput of MiniMax's official H3 endpoint and an average 15x speedup over models of comparable quality in its tests.
- Design Arena ranks H3 Max first on its recent image-to-video leaderboard at 1,341 Elo, ahead of official MiniMax H3 at 1,333, according to fal's announcement.
- Artificial Analysis ranks fal's H3 first on its image-to-video-with-audio leaderboard at 1,201 Elo across 2,177 samples, ahead of Seedance 2.0, MiniMax H3, Veo 3.1, and Kling 3.0.
- fal says it post-trained H3 on substantial new data with an in-house reinforcement-learning framework, while its inference team co-designed and optimized the serving stack alongside the evolving model.