In a remarkable achievement that highlights China’s rapid progress in artificial intelligence, Moonshot AI has unveiled its new model, Kimi K2 Thinking, which has outperformed leading American AI systems like ChatGPT and Claude in several key benchmark tests — marking a new era of East–West competition in the AI world.
Chinese tech company Moonshot AI has announced that its latest model, Kimi K2 Thinking, has outshined leading global AI systems by achieving outstanding results in multiple benchmark evaluations.
In the Humanity’s Last Exam (HLE) benchmark, Kimi scored 44.9%, surpassing GPT-5 (High) at 41.7% and Claude Sonnet 4.5 Thinking, which scored 32%.
The Chinese model also led in the BrowseComp test — designed to measure self-directed browsing and research capabilities — achieving 60.2%, compared to 54.9% for GPT-5 and 24.1% for Claude.
However, in the field of competitive programming, ChatGPT remains ahead, with Kimi K2 scoring 83.1% on the LiveCodeBench V6 test, versus 87% for GPT-5.
Notably, Kimi K2 Thinking stands out as one of the few advanced AI models available completely free and without message limits through its official website and apps on both Apple and Android stores — allowing users to explore its advanced capabilities firsthand.
Analysts view this as a turning point for China’s AI industry, signaling that the country is no longer merely catching up with Western competitors but is now surpassing them in certain areas — ushering in a new phase of global competition for technological dominance.
