Moonshot AI releases the most powerful open-source reasoning model, Kimi K2 Thinking, with intelligent inference capabilities surpassing GPT-5
Jinse Finance reported that Moonshot AI announced the release of Kimi K2 Thinking, their most powerful open-source reasoning model. As a reasoning model, Kimi K2 Thinking performs step-by-step inference while using tools, achieving state-of-the-art results in HLE, BrowseComp, and other benchmark tests, and has made significant progress in reasoning, agent search, coding, writing, and general capabilities. Kimi K2 Thinking can execute up to 200 - 300 consecutive tool calls without human intervention, conducting coherent reasoning across hundreds of steps to solve complex problems. In the Human-Level Exam (HLE), an extremely challenging benchmark, Kimi K2 Thinking scored 44.9%, surpassing advanced models such as Grok4, GPT-5, and Claude 4.5. If using Kimi K2 Thinking Heavy, the score can further reach 51%.
Disclaimer: The content of this article solely reflects the author's opinion and does not represent the platform in any capacity. This article is not intended to serve as a reference for making investment decisions.
You may also like

JPMorgan acquires stake in leading Ethereum reserve Bitmine, holding position valued at $102 million
Publicly listed company Falconedge spends $1.6 million to complete its first purchase of approximately 15 BTC
