On September 5, Moonshot AI officially announced the release of the latest version of Kimi K2 model (0905), which further optimizes programming capabilities and task processing performance.
According to the introduction, the new version of Kimi K2 model has significantly improved its Agentic Coding capability, showing better performance in both public benchmark tests and real programming tasks. At the same time, its front-end programming experience has been upgraded, making the code more aesthetically pleasing and practical.
In terms of model capabilities, the context length of K2 has been expanded from 128K to 256K, supporting more complex and long-term task requirements. In addition, the new version introduces a high-speed API, with an output speed of 60-100 Token/s, greatly improving response efficiency.
Moonshot AI stated that this upgrade aims to make Kimi K2 better serve developers' application needs in diverse scenarios.
