Hao AI Lab 发布 FastVideo FastH3，基于 MiniMax H3
(13/13) FastVideo FastH3 builds on @MiniMax_AI H3. We thank the MiniMax team for releasing its weights and code.
UCSD Hao AI Lab 发布 FastVideo FastH3，基于 MiniMax H3 构建，并感谢 MiniMax 开源其权重与代码。在 8× GB200 上，5 秒 1344×768 带音频片段 3.1 秒输出成品 MP4，10 秒片段 6.3 秒（NVFP4 下 5.4 秒），负载下每天约 11,000 条视频，8× RTX PRO 6000 上约 3,400 条。UniServe 还作为 NVIDIA Dynamo 的实验后端，通过 uniserve-dynamo-worker 嵌入其 Rust 调度器与引擎。
正文
当前显示原文，中文全文译文尚未就绪。
(13/13) FastVideo FastH3 builds on @MiniMax_AI H3. We thank the MiniMax team for releasing its weights and code.
We thank @nuvalab and @reactorworld for their collaboration, feedback and insights during UniServe’s development, and Nuva Lab for access to the production hardware used in our benchmarks.
We thank Guan Luo, Qi Wang, Ryan McCormick, Harry Kim, and the rest of the NVIDIA Dynamo team for their support in integrating Dynamo with UniServe, bringing UniServe closer to a production-grade serving engine.
We also thank @NVIDIAAI FastGen (Julius Berner, Chao Liu, Arash Vahdat), and @NVIDIAAI Enterprise Products (Pengcheng Li, Cliff Woolley) team for collaborating on FastH3.
Finally, we are extremely grateful to @vllm_project , @NVIDIAAI and @mbzuai for their continued sponsorship and support of FastVideo.