Kimi Code

What it does

Kimi Code takes plain-language instructions plus the project files on your machine and acts on them. The agent can read and search code, edit files, run commands such as tests and builds, check the results, and search the web or read web pages. Output is code changes, command results and written explanations inside a chat-style session.

You can use it through a Desktop app (macOS, Windows, Linux), a command-line tool, or a VS Code extension, signing in with a Kimi account rather than managing keys. Other editors such as JetBrains and Zed can connect through the CLI's ACP protocol. Members can also create an API key and point third-party agent tools, including Claude Code, OpenCode and Codex, at its OpenAI- or Anthropic-compatible endpoints. It offers models with up to 1M-token context and faster, higher-quota variants.

It suits developers who work on multi-file engineering tasks and want one membership across several tools. Access requires a Kimi membership; model availability, context size and speed options depend on the tier. Usage is rate-limited under the subscription, and the docs point teams needing pay-as-you-go billing or product integration to the separate Kimi Platform. Altering a tool's client identifier is treated as a violation.

What it can do

• Read, search and analyze an existing codebase
• Write and edit code, fix bugs and refactor across files
• Run tests, builds and scripts, then verify results
• Search the web and read web page content
• Use models with up to 1M-token context, with regular and high-speed versions
• Provide an API key for third-party coding tools via OpenAI- and Anthropic-compatible endpoints

Works with

• Claude Code
• OpenCode
• Codex
• VS Code
• JetBrains and Zed (via the CLI's ACP protocol)

Example workflows

Fix a failing test from the terminal

Input: a local repository and a plain-language bug description. Steps: install the CLI, run kimi, log in with /login, describe the failure and let the agent read the code, edit files and run the tests. Output: code edits and test results to review.

Use Kimi Code models in another agent tool

Input: a Kimi membership and an API key from the Kimi Code Console. Steps: create the key, set the Base URL and key in Claude Code's environment variables, choose a model ID such as k3.1-pro. Output: the third-party tool running on Kimi Code models.

Before you choose

• Model access, 1M-context unlocking and high-speed versions depend on membership tier, and high-speed use consumes more quota.
• The subscription is rate-limited; pay-as-you-go API use belongs to the separate Kimi Platform.
• The CLI was rebuilt on Node.js and the legacy Python version is no longer maintained.

Listed under

• Coding
• Code review and testing
• Coding agents
• AI coding
• CLI
• Code editing
• Coding agent
• Desktop app

Get started

Needs a Kimi membership; sign in with your Kimi account in the client.

1. Get a Kimi membership plan that includes Kimi Code.
2. Install the Desktop app, run the CLI install command, or add the Kimi Code VS Code extension.
3. Sign in with your Kimi account, then open a project and give the agent a task.

Billed as a Kimi membership subscription, monthly or annual, with rate limiting. Plan prices were not confirmed; see the Kimi pricing page.

We couldn’t find published prices for Kimi Code. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Kimi Code - AI coding agent powered by Kimi K3.1 Pro, with CLI, VS Code extension, Desktop App, and more ↗www.kimi.com
• Kimi Code Overview | Kimi Code Docs ↗www.kimi.com

Company website

kimi.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.