Kimi K2 Thinking
Moonshot AI · text → text
Kimi K2 Thinking is the latest generation open-source reasoning model, designed for advanced multi-step thinking and dynamic tool invocation. It achieves state-of-the-art performance on benchmarks such as HLE and BrowseComp, demonstrating exceptional reasoning depth and stability. Built with native INT4 quantization and a 256k context window, Kimi K2 delivers high precision with significantly reduced inference latency and memory usage, supporting 200–300 consecutive tool calls for long-context reasoning, complex task planning, and intelligent agent applications.
Input$0.548 /M
Output$2.192 /M