MiniMax Launches Open-Weights Video Model H3 Same Day as ByteDance's Closed Seedance 2.5
AI video model releases
MiniMax Launches Open-Weights Video Model H3 Same Day as ByteDance's Closed Seedance 2.5
Chinese startup MiniMax released its video generation model H3 on July 31 under open weights, supporting text-to-video and reference-to-video generation.
ByteDance launched its closed-source Seedance 2.5 on the same day, setting up a direct comparison between open-weights and closed-API approaches in China's video model market.
Inference platform fal already has H3 API endpoints live, letting developers either self-deploy the open weights or call the hosted service directly.
No independent third-party evaluation data yet confirms how H3 compares to Seedance 2.5 on metrics like long-form generation, complex motion, or cross-frame consistency, only limited community sample comparisons on X exist.
H3's specific license terms, including whether commercial use is permitted or usage limits apply, have not been confirmed despite the "open weights" label.
cho.sh