PANews December 16 news, according to the WeChat public account 'Dark Side of the Moon Kimi', today, Kimi released the visual thinking model k1. The k1 model is built on reinforcement learning technology, natively supports end-to-end image understanding and thinking chain technology, and expands its capabilities beyond mathematics into more fundamental scientific fields. In benchmark capability tests across fundamental scientific disciplines such as mathematics, physics, and chemistry, the first-generation k1 model outperformed the global benchmark models OpenAI o1, GPT-4o, and Claude 3.5 Sonnet.