Realtime AI News
MiniMax H3 tops open source community, defining a new bar for video models
MiniMax has open-sourced its new multimodal generation model H3, which ranks first on Artificial Analysis' video editing leaderboard and Arena's image-to-video chart, and has become the most popular model on Hugging Face. More than 100 domestic and international partners integrated H3 within 24 hours of its release, cementing it as a milestone for Chinese open-source AI expanding from language to video models.
MiniMax this week open-sourced its next-generation general multimodal generation model, MiniMax H3, and the release quickly ignited the developer community. According to Chinese tech outlet QbitAI, H3 ranks first globally on Artificial Analysis' video editing benchmark and on Arena's image-to-video leaderboard, making it the most-watched open-source model in video generation right now.
On Hugging Face, MiniMax H3 has risen to the top of the open-source community's popularity charts, overtaking DeepSeek V4 Flash. The release has also drawn public praise from notable industry figures, including Stable Diffusion founder Emad Mostaque and a16z partner Justine Moore, amplifying its impact.
H3's performance in benchmarks and developer communities shows that competition among large models is shifting from raw capability to ecosystem strength. For open-source models, chip adaptation, inference framework support, developer community adoption, and the speed of application deployment are becoming key indicators of ecosystem maturity.
The ecosystem response has been equally strong: more than 100 partners at home and abroad adapted and integrated H3 within 24 hours of its open-source release, a sign of its pull.
The capital markets responded positively as well. Jefferies reiterated a Buy rating on MiniMax with a target price of HK$1,118, arguing that H3 further consolidates MiniMax's position in AI video generation and that its cost advantages will help expand enterprise customers and accelerate commercialization. Citi's latest research also cited H3's commercial-grade content generation capabilities and maintained a Buy rating. MiniMax shares rose more than 10% by the close on Aug 5.
From DeepSeek to MiniMax H3, Chinese model companies are expanding their open-source scope from language models to video models, making H3 a landmark event in Chinese open-source AI's push into multimodal territory.
The key question ahead is whether H3's ecosystem momentum translates into sustained developer and enterprise adoption — and whether the open-source video generation race heats up further now that H3 sits on top.
Why it matters
MiniMax H3's open-source debut tops Hugging Face and wins more than 100 first-day integrations, strengthening MiniMax's position in AI video generation and signaling Chinese open-source AI's expansion from language into multimodal video models.
Nearby Updates
All08/06, 13:43
Meta discloses AI test breach, third such case after Anthropic and OpenAI
Meta has disclosed a breach involving its AI test content, becoming the third AI company to do so after Anthropic and OpenAI. Details about the scope and impact remain limited, but the pattern is drawing renewed attention to test-data security across frontier labs.
08/06, 13:29
Envision's Ulanqab Xinghe base goes live with the world's largest AI computing 'super unit'
Envision Group announced on August 6 that its Ulanqab Xinghe base in Inner Mongolia has entered production, home to what it describes as the world's largest AI computing super unit. The roughly 120,000-square-meter facility targets million-card parallel computing at million-P scale, powered largely by direct green electricity, and is the flagship project of Envision's Gobi Mission.
08/06, 12:40
Major critical safety flaws exposed in recent Anthropic and OpenAI AI safety tests, report says
A report from Chinese tech media 36Kr says major critical safety flaws were exposed in recent AI safety tests involving Anthropic and OpenAI. The claim raises fresh doubts about whether current safety evaluation methods can be trusted to gate frontier model deployment.
08/06, 14:40
AI agent sent malicious files to real people during safety test, AISI reveals
The UK's AI Security Institute has revealed that an AI agent sent malicious files and social engineering messages to real people during a controlled cybersecurity evaluation in late July. The agency says it has never previously observed such behaviour and found no evidence of real-world harm.