Chinese artificial intelligence firm Moonshot AI has released a new reasoning variant of its open-source Kimi K2 model that has outperformed OpenAI’s GPT-5 and Anthropic’s Claude Sonnet 4.5 in several benchmarks, eliciting high praise from industry experts on social media.
Advertisement
The Beijing-based start-up on Thursday said Kimi K2 Thinking was now accessible via Kimi.com and the model’s application programming interface (API), allowing developers and enterprises to use the model’s functionalities into their applications.
In a blog post on online developer platform GitHub, Kimi researchers said Kimi K2 Thinking set “new records across benchmarks that assess reasoning, coding and agent capabilities”.
Those results reflected how Chinese AI companies have closed the performance gap between their open-source models and US peers’ closed-source models. Previously, new Chinese open-source AI models achieved international popularity, but were behind US closed-source models in terms of performance.
“Today is a turning point in AI. A Chinese open-source model is #1,” wrote Deedy Das, a partner at early stage venture capital firm Menlo Ventures, in a Friday post on X. “Seminal moment in AI.”
Advertisement
Kimi K2 Thinking outperformed closed-source models GPT-5 and Claude Sonnet 4.5 with a score of 44.9 per cent on Humanity’s Last Exam, a large language model (LLM) benchmark consisting of 2,500 questions across a broad range of subjects, according to the GitHub blog post.