Skip to main content
Chinese AI Tools
ProductsModelsIntegrationsRankingsLatest changes
TopicsAvailability TrackerUse casesSubmit a toolAccount
ZHSearch

Chinese AI Tools

Independent directory for Chinese AI products. Product availability, pricing and terms can change. Verify before commercial use.

Editorial standardsClaim productUpdate infoGet featuredAdvertise

Moonshot AI

Kimi Code

Moonshot AI's MIT-licensed terminal coding agent and member coding backend for Claude Code, Roo Code and other compatible agents.

Globally availableFull English UILimited APIFreemium

Quick answer

The public repository confirms that Kimi Code is an installable MIT-licensed agent platform, while the K2.7 Code resource confirms its current default hosted coding model.

Official siteDocumentation
Pricing
MIT-licensed CLI; annual-billing Kimi Code plans are listed at $15, $31, $79 and $159 per month, with model usage also possible through API billing or configured providers
Availability
Globally available
API
Limited API
Last checked
2026-08-17
Pricing detailsUse-case fitSource evidence

Kimi Code CLI is Moonshot AI's open-source terminal coding agent. It can read and edit code, run shell commands, search files, fetch web pages and choose next actions from tool feedback. The single-binary CLI works with Kimi Code OAuth or a Moonshot AI Open Platform API key and can also use compatible providers. Kimi Code membership benefits can additionally serve third-party tools: the official guide configures Claude Code through an Anthropic-compatible endpoint and Roo Code through an OpenAI-compatible endpoint, while also naming OpenCode, OpenClaw and Hermes as compatible paths. Kimi K2.7 Code is the default thinking model; third-party calls require Thinking mode to reach K2.7 Code, otherwise they route to K2.6. The broader CLI feature set includes video input, conversational MCP configuration, a plugin marketplace with surfaced trust levels, isolated coder/explore/plan subagents, lifecycle hooks for approval and auditing, and Agent Client Protocol integration for Zed, JetBrains and other ACP clients.

Topic guide

Kimi models, API and agent workspace

A source-backed map of Kimi K3, K2.7 Code, the Kimi API, Kimi Code, Kimi Claw, WebBridge and productivity creation surfaces.

Editorial verdict

Best for

Developers who want a fast open terminal agent with multimodal coding context, parallel subagents and editor integration.

Avoid if

Avoid enabling broad command, plugin or repository access before defining approvals, hooks, audit logs and provider data boundaries.

Why it matters

The public repository confirms that Kimi Code is an installable MIT-licensed agent platform, while the K2.7 Code resource confirms its current default hosted coding model.

Trust: 7/7 sources verified, recently checkedCoverage: 100/100

Pricing

MIT-licensed CLI; annual-billing Kimi Code plans are listed at $15, $31, $79 and $159 per month, with model usage also possible through API billing or configured providers

Payment

Free open-source CLI, Kimi Code OAuth plans, Moonshot AI API billing, Compatible provider billing

Commercial use

Commercial use should follow the current product, API, model license and billing terms.

Privacy

Review prompt, file, media upload, retention and training-use terms before sensitive workloads.

Use-case fit

Terminal software engineering

Strong

Use the TUI for repository exploration, code changes, shell commands and iterative tool-driven implementation.

K2.7 Code hosted coding

Strong

Use Kimi Code when you want K2.7 Code as a productized terminal or IDE coding model rather than raw API access.

Third-party coding-agent backend

Strong

Use the documented Anthropic-compatible route for Claude Code or the OpenAI-compatible route for Roo Code; the guide also names OpenCode, OpenClaw and Hermes.

Video-to-code workflows

Strong

Attach a screen recording or demo when visual behavior is easier to show than describe.

Parallel agent software tasks

Strong

Dispatch coder, explore and plan subagents into isolated contexts while preserving the main conversation.

ACP editor integration

Strong

Run `kimi acp` so Zed, JetBrains or another ACP client can drive the logged-in CLI session.

Browser-assisted local agent work

Medium

Pair with Kimi WebBridge when a local coding agent needs Chrome or Edge automation with existing login sessions.

Global user checklist

RegistrationConfirmedThe CLI supports Kimi Code OAuth or a Moonshot AI Open Platform API key. Third-party agent access requires Kimi membership activation and a Kimi Code Console API key.
English UIConfirmedThe repository and documentation provide English and Chinese content.
API and docsConfirmedDocs cover the CLI plus Claude Code and Roo Code setup. Roo Code is documented with a 262,144-token context window, 32,768 maximum output tokens, streaming, image support and medium reasoning effort.
International paymentConfirmedThe K2.7 Code resource lists Kimi Code annual-billing plan prices for Moderato, Allegretto, Allegro and Vivace.
Commercial useConfirmedThe CLI source is released under the MIT License; model and hosted-service usage remains subject to provider terms.
Data and privacy termsReviewReview what code, videos, web content and tool output are sent to the selected model provider.

Checked against stable repository version 0.36.1 on August 17, 2026. Newer main-branch work is not treated as released; CLI features, plans, endpoints and provider compatibility can still change quickly.

Pros

  • - MIT-licensed source with macOS, Linux and Windows single-binary installers
  • - Kimi K2.7 Code is now the default model with thinking enabled by default
  • - Official setup guides cover Claude Code and Roo Code through compatible API protocols
  • - Video input can turn screen recordings and demonstrations into coding context
  • - Conversational MCP configuration and plugins reduce manual integration setup
  • - Coder, explore and plan subagents can run focused tasks in isolated contexts
  • - ACP connects the same CLI session to Zed, JetBrains and compatible editors
  • - Stable 0.36.1 adds plan and goal views, background-task inspection, subagent status and improved long-session reliability

Cons

  • - Repository, shell, network and plugin permissions need explicit team policy before rollout
  • - OAuth plans, API billing and third-party provider costs are separate from the free CLI
  • - K2.7 Code does not support non-thinking mode, so non-thinking Kimi Code requests are served by K2.6
  • - Third-party agent access requires an active Kimi membership and a separately created Kimi Code API key
  • - Plugin trust labels help discovery but do not replace source and permission review

Decision paths

mimocode

qwen-code

opencode-go

codebuddy

Sources

Kimi Code GitHub repository

Official · EN · verified 2026-08-04

Documents the MIT-licensed CLI, installation, video input, MCP, plugins, subagents, hooks, ACP and provider support.

Kimi K2.7 Code official resource

Official · EN · verified 2026-06-17

Confirms K2.7 Code as the Kimi Code default model and lists current annual-billing Code plan prices.

Kimi Code third-party coding-agent guide

Documentation · EN · verified 2026-06-20

Documents membership prerequisites, API-key creation, Claude Code and Roo Code configuration, compatible endpoints, context limits and K2.7 thinking-mode routing.

Kimi Code 0.36.1 release

Official · EN · verified 2026-08-17

Confirms the current stable release, plan and goal controls, background-task and subagent views, command search and reliability fixes.

Kimi Code documentation

Documentation · EN · verified 2026-06-13

Official documentation for setup, interaction, sessions, IDEs, configuration and commands.

Kimi Code official page

Official · EN · verified 2026-06-13

Confirms the hosted Kimi Code product and plan route alongside the open CLI.

Kimi WebBridge help center

Documentation · EN · verified 2026-05-17

Confirms WebBridge can pair with Kimi Code and other local agents.

Last checked: 2026-08-17

Reviews

No approved reviews yet.

Availability snapshot

Availability
Globally available
English UI
Full English UI
API
Limited API
Rating
4.7 (0)

Latest updates

Latest changes
Release · 2026-08-14

Kimi Code 0.36.1 improves long-running work controls

Kimi Code 0.36.1 adds plan and goal indicators, a plan viewer, background Bash task filtering and details, subagent status views, fuzzy slash-command search with pinyin support and automatic session titles. The stable release also fixes self-hosted OpenAI-compatible approval history, Windows file watching, MCP OAuth cancellation, large session export and long-session forking.

Open source · 2026-08-04

Kimi K3 full weights are now public

Moonshot AI published Kimi K3's full weights, model code and custom model license on Hugging Face on July 27. The public repository contains an MXFP4-compressed checkpoint split across 96 weight shards, closing the launch announcement's open-weight commitment while leaving substantial infrastructure and license-review requirements for self-hosting.

Release · 2026-07-17

Moonshot launches Kimi K3

Moonshot AI launched Kimi K3 as a hosted 2.8T-parameter, native-vision MoE model with a 1M-token context window. It became available in Kimi, Kimi Work, Kimi Code and the Kimi API as kimi-k3 at $0.30/M cache-hit input tokens, $3/M cache-miss input tokens and $15/M output tokens. At launch, Moonshot scheduled the full-weight release for July 27; that release is now tracked separately. Published limitations include sensitivity to dropped thinking history and possible over-proactiveness on ambiguous tasks.

API · 2026-06-21

Kimi K2.7 Code API contract documented

Kimi's K2.7 Code quickstart documents the kimi-k2.7-code and kimi-k2.7-code-highspeed model IDs, a 256K context window and a default 32K maximum output. HighSpeed targets about 180 tokens per second and up to 260 tokens per second for short contexts, though capacity is currently limited. Thinking cannot be disabled; temperature, top_p, n and penalty values are fixed, tool_choice only accepts auto or none, and multi-step tool calls must preserve reasoning_content. The OpenAI-compatible API also supports image and video content in multimodal tool results.

Submit a review