China’s MiniMax Launches New Flagship AI Model M2.5
Chinese AI models offering strong coding and agentic capabilities are gaining worldwide recognition from app developers.
Chinese AI developer MiniMax launched its latest flagship open-source large language model, M2.5, adding to the recent flood of new model releases from China.
Shanghai-based MiniMax said M2.5 offers strong performances in coding and agentic tasks, at significantly lower prices than major U.S. proprietary models. The model, like other recently launched Chinese models, is likely to appeal to application developers who are seeking decent but cheaper alternatives to Anthropic’s Claude and other U.S. models.
MiniMax went public in Hong Kong last month. The company’s stock was up nearly 10% as of Friday morning after its announcement of M2.5.
Chinese AI models are gaining worldwide recognition. Beijing-based Zhipu’s latest flagship model GLM-5, also released this week, has received positive reviews, especially in coding. Earlier this month, another Chinese competitor, Beijing-based Moonshot AI, launched its new model called Kimi K2.5, which last week became the most popular model by usage on LLM marketplace OpenRouter.