MiniMax H3 Tops Open-Source Community, Setting a New Benchmark in Video AI
MiniMax
DeepSeek
MiniMax has open-sourced its new multimodal generation model MiniMax H3, which now ranks first globally in AI video editing and image-to-video leaderboards, and has become the most popular model on Hugging Face. The model's rapid adoption and strong market performance highlight a shift in AI competition from raw capability to ecosystem readiness.
This week, MiniMax officially open-sourced its new-generation general multimodal generation model MiniMax H3, which has topped both the Artificial Analysis video editing benchmark and the Arena image-to-video leaderboard, with praise from Stable Diffusion founder Emad Mostaque and a16z partner Justine Moore. On Hugging Face, H3 has become the most popular model, surpassing DeepSeek V4 Flash. Within 24 hours of release, over a hundred domestic and international partners had adapted and integrated H3. Analysts at Jefferies reiterated a Buy rating with a target price of HK$1118, noting H3 strengthens MiniMax's position in AI video generation and its cost advantage will help expand enterprise clients. Citi also maintained a Buy rating, citing H3's commercial-grade content generation capability. By the close on the 5th, MiniMax shares had risen over 10%. The release marks a significant step for Chinese open-source AI expanding from language to video models.
Source: QbitAI 量子位 —
original
