According to TVB News, four of the top five AI models by usage volume during the week of February 17-23 were from China: MiniMax M2.5, Kimi K2.5 (Moonshot AI), GLM-5 (Zhipu AI), and DeepSeek V3.2. This marks the first time Chinese AI models have surpassed US usage rankings globally. The data from OpenRouter, the world's largest AI model integration platform, represents a significant shift in the competitive landscape as Chinese companies establish their position in the AI market. As reported by CNBC, the MiniMax M2.5 model (launched February 13) has attracted significant developer traction, reaching one-third of Anthropic Claude's usage on OpenRouter. This represents a substantial achievement as the Chinese model competes directly with established US alternatives.
UBS analysts have reiterated Tencent as their industry top pick with a Buy rating and a target price of HKD780, maintaining optimism about the company's massive potential in the agentic AI domain. According to a UBS report, Tencent's stock has dropped around 13% YTD, underperforming the HSTECH and HSI over the same period, primarily due to investor concerns about potential VAT increases for tech companies and heightened regulatory scrutiny. However, UBS believes the key to success in AI lies in integrating high-quality models, a strong user base, and extensive data. With 1.4 billion users across WeChat's diverse application scenarios, UBS remains confident that Tencent's unique advantages are difficult to replicate in the competitive AI landscape.
UBS analysts expressed strong preference for MiniMax among China's emerging AI models, citing superior performance-to-cost ratios. The MiniMax M2.5 model (launched February 13) has attracted significant developer traction, reaching one-third of Anthropic Claude's usage on OpenRouter. This represents a substantial achievement as the Chinese model competes directly with established US alternatives. The MiniMax M2.5 model is priced at one-tenth the cost of Claude Opus 4.6, yet rivals it in performance capabilities, driving significant developer adoption across various sectors. The company in mid-February released its M2.5 model, with performance rivalling Claude's latest Opus 4.6 at a lower price point.
The competitive pricing strategy is proving effective in attracting developer communities. The MiniMax M2.5 model's affordability is enabling widespread adoption among developers seeking cost-effective AI solutions. This developer adoption is crucial for building ecosystems and expanding the practical applications of AI technology. The MiniMax M2.5 model's one-third usage share on OpenRouter demonstrates the significant market penetration achieved by the Chinese AI model. As reported by CNBC TV18, the MiniMax M2.5 model's affordability is enabling widespread adoption among developers seeking cost-effective AI solutions.
The release of multiple AI models by Chinese companies indicates a strategic approach to market penetration. This competitive landscape suggests companies are positioning themselves across different price points and performance levels to capture various market segments. The timing coincides with growing speculation about potential updates to existing AI models and represents a coordinated effort to establish Chinese AI dominance in the global market. As speculation over a new version of the DeepSeek AI model grows, several other Chinese tech companies have released their own generative artificial intelligence models in the past few weeks, including Alibaba's Qwen 3.5 and ByteDance's video-generating Seedance 2.0.