Kimi CLI v1.37.0 Changelog

This update is all about making the Kimi command-line tool (the text-based way you control the AI) smarter and less frustrating to use. It mostly matters to developers and power users who rely on this tool to help write code or automate tasks, giving them a smoother experience with better model management and fewer annoying glitches.

What's new

• Automatic model updates: When you open the tool, it automatically refreshes its list of "managed models" (the specific AI brains it can use). You no longer have to manually update this list to see the newest options.
• Friendlier model names: The tool now surfaces the API display_name. This means instead of showing you a confusing internal code name for an AI model, it shows the actual, readable name the creators gave it.

Improvements

• Clearer settings instructions: The documentation (the instruction manual) was updated to clarify how to format "quoted TOML keys." TOML is just a type of settings file. If you want to use a model with a dot in its name (like model.name), the instructions now clearly explain that you need to put it in quotes so the tool doesn't get confused.

Fixes

• Fixed the AI freezing during background tasks: There was a bug in the "agent loop" (the cycle the AI uses to think and act). If the AI asked a background task to run—like searching a file or downloading something—the main AI loop would accidentally shut down before the task finished. This update keeps the loop alive and waiting so the AI can actually use the results of those background tasks.

Breaking changes

No breaking changes in this release.