On September 2, Moonshot AI announced that Kimi API had completed native support for OpenAI’s Responses API and Anthropic’s Messages API. Developers using Codex or Claude Code can now connect directly to multiple Kimi coding models through a custom model provider, without requiring additional format conversion or a local proxy.
Breaking Down Barriers: Kimi API Natively Supports Protocols Used by Two Major Coding Tools
The update allows developers to call Kimi models directly from two of the most widely used AI coding tools. Supported models include kimi-k3, kimi-k2.7-code-highspeed, kimi-k2.7-code, and kimi-k2.6, covering a range of coding scenarios from high-performance development to high-speed inference.
Moonshot AI has also provided detailed integration guides covering the configuration processes for Codex CLI, the Codex/ChatGPT desktop client, and Claude Code. Developers can complete the integration by setting the base_url to https://api.moonshot.cn/v1 for Codex and https://api.moonshot.cn/anthropic for Claude Code.
Streamlined Setup: One-Click Configuration Simplifies Developer Workflows
For users who already have local agents such as Kimi Code or Kimi Work installed, they can simply send instructions to the agent and have it follow the official documentation to complete the configuration automatically, further lowering the barrier to adoption.
It is worth noting that OpenAI’s Responses API currently supports text and image inputs, but does not yet support video input. The official documentation recommends using ffmpeg to extract keyframes from a video before submitting them to the model for analysis.
In addition, the model selector in the Codex desktop client, now integrated into ChatGPT following an update, may display the option as “Custom.” However, the model actually used will be the Kimi model configured by the user.
Industry Outlook: Interoperability Among AI Coding Tools Becomes a New Competitive Focus
According to IT Home, the native integration eliminates the previously cumbersome process of protocol conversion or deploying a local proxy between developers and Kimi API. TMTPost noted that the move significantly lowers the barrier for developers looking to integrate Kimi API into AI coding workflows, potentially improving coding efficiency.
As AI-assisted coding tools become increasingly widespread, interoperability between models is emerging as a growing industry focus. By supporting the protocols of two major AI coding tools, Kimi API is aligning with this trend while giving developers greater flexibility in choosing the models they use.