Kimi Code CLI is Moonshot AI’s terminal coding agent, and it is the first big-model-vendor harness whose default path is a challenger model instead of Claude or GPT.
What it is #
A terminal AI coding agent that reads and edits code, runs shell commands, searches files, fetches pages, and picks its next step from the feedback.
It works out of the box with Moonshot’s Kimi models and can be pointed at other compatible providers.
Distribution is a single binary (install script or npm), with a purpose-built TUI, video input in chat, conversational /mcp-config editing, a plugin marketplace with per-install trust levels, built-in coder, explore, and plan subagents, lifecycle hooks, and ACP support so Zed or JetBrains can drive a session.
The code lives in two repositories: the original MoonshotAI/kimi-cli (Apache-2.0, created 2025-10-15, archived on GitHub as of 2026-10-02) and the newer canonical MoonshotAI/kimi-code (MIT, created 2026-05-22).
Status #
Active and big-vendor backed, about 19.2k combined stars as of 2026-10-04: kimi-code at 7,767 stars, kimi-cli at 11,435 stars, with the older repository now archived by Moonshot (its last push was 2026-09-22).
The independent FrontierHarness Eval pegs it mid-pack on quality and cheap on cost: 56.7 percent pass at a $3.65 median cost per task, with an 88 percent cache rate on Kimi K3.
The Hacker News footprint is thin (a 5-point 2025-10-31 thread and a 1-point 2026-01-27 one), which I read as distribution through Kimi’s own channels rather than developer-mindshare gravity.
Strengths #
- Cheapest credible big-vendor harness path: the same token budget that buys one Claude Code task in FrontierHarness bought several Kimi Code ones.
- Full agent surface in one binary: subagents, hooks, marketplace skills, MCP, and ACP editor integration without plugins or Node.js.
- The model and the harness come from the same vendor, so prompt-to-model tuning is first-party.
Cautions #
- The repository split (Apache-2.0
kimi-cliversus MITkimi-code) has resolved towardkimi-code: the official docs state the CLI was rebuilt on Node.js and that the old Python version will no longer be maintained, with a migration guide pointing at the new codebase (docs, as of 2026-09-12). The wind-down is now complete: GitHub shows the
kimi-clirepository archived as of 2026-10-02, after its final 1.52.0 release on September 22, 2026. - Your tokens flow to Moonshot’s API in the default path, the same vendor-dependency trade Claude Code and Codex make, just with a cheaper vendor.
- Thin independent coverage means the FrontierHarness numbers are nearly the only third-party evidence available.
Pricing #
The harness is free and open source.
Usage is Kimi Code OAuth (your Kimi subscription) or a Moonshot Open Platform API key: kimi-k3 at $0.72/$13.00 per million tokens (cache reads listed at $0.70) and kimi-k2.7-code at $0.67/$3.35 (cache hits $0.18) as of 2026-10-04 (verified via OpenRouter’s model list; Moonshot’s CN platform listed the same models at ¥20/¥100 and ¥6.50/¥27 as of 2026-09-09), with other OpenAI-compatible providers configurable.
Price history #
| Date | Plan | Change | Source |
|---|---|---|---|
| 2026-09-09 | Moonshot CN platform | kimi-k3 at ¥20/¥100 and kimi-k2.7-code at ¥6.50/¥27 per million tokens; harness free and open source. | Moonshot CN platform listing, see References |
| 2026-09-20 | API (OpenRouter) | kimi-k3 at $1.70/$8.50 per million tokens (cache hits $0.17), kimi-k2.7-code at $0.71/$3.21. | |
| 2026-09-22 | API (OpenRouter) | kimi-k3 raised to $3.00/$15.00 per million tokens (cache hits $0.30); kimi-k2.7-code completion raised to $3.30 (cache hits $0.18). | |
| 2026-09-24 | API (OpenRouter) | kimi-k2.7-code input dropped to $0.66 per million tokens (completion $3.30 and cache $0.18 unchanged); kimi-k3 unchanged at $3.00/$15.00. | |
| 2026-10-02 | API (OpenRouter) | kimi-k3 dropped sharply to $0.4255/$10.00 per million tokens (cache reads listed at $0.4255, equal to input); kimi-k2.7-code at $0.67/$3.35 (cache hits $0.18). | |
| 2026-10-02 | API (OpenRouter) | kimi-k3 raised again to $2.70/$13.50 per million tokens (cache reads $0.27); kimi-k2.7-code unchanged at $0.67/$3.35 (cache hits $0.18). | |
| 2026-10-04 | API (OpenRouter) | kimi-k3 dropped to $0.72/$13.00 per million tokens (cache reads $0.70); kimi-k2.7-code unchanged at $0.67/$3.35 (cache hits $0.18). |
Compared to #
- Codex: the other big-vendor open CLI, stronger on FrontierHarness quality (66.7 percent) but no longer on price, at $3.47 versus $3.65 median cost per task as of 2026-09-13.
- Qwen Code: Alibaba’s fork-and-rebrand path into the same market; Kimi Code is a from-scratch Moonshot product rather than a Gemini CLI derivative.
- OpenCode: the provider-neutral alternative when you want the harness decoupled from any one model vendor.
Bottom line #
Recommended for engineers already inside the Kimi ecosystem and for token-payers who want a big-vendor-polished agent on challenger-model economics. Not for teams that need local models, or anyone who needs more than one independent evaluation before trusting a harness.
Changes #
- 2026-09-05 - Created from the same-day entrant resolution as Moonshot’s MIT CLI.
- 2026-09-06 - Recorded the docs confirming the Node.js rebuild with the Python version unmaintained, and replaced the measured-cost comparison.
- 2026-09-16 - Refreshed star counts (about 18.8k combined) and updated the kimi-k2.7-code output price to $3.21 per million tokens on OpenRouter.
- 2026-09-18 - Refreshed star counts (about 18.9k combined) and recorded the kimi-k3 price drop on OpenRouter to $2.10/$10.95 per million tokens with cache hits at $0.23.
- 2026-09-20 - Recorded the further kimi-k3 price drop on OpenRouter to $1.70/$8.50 per million tokens with cache hits at $0.17.
- 2026-09-20 - Added the Price history section tracking price changes in a table, per the new owner rule.
- 2026-09-22 - Recorded the kimi-k3 price rise on OpenRouter to $3.00/$15.00 per million tokens (cache hits $0.30) and the kimi-k2.7-code completion rise to $3.30 (cache hits $0.18).
- 2026-09-22 - Noted kimi-cli shipping 1.52.0 on September 22 despite the migration notice, and refreshed repository counters.
- 2026-09-24 - Recorded the kimi-k2.7-code input price drop on OpenRouter to $0.66 per million tokens (completion and cache unchanged, kimi-k3 unchanged) and refreshed repository counters.
- 2026-10-02 - Recorded Moonshot archiving the original
kimi-clirepository on GitHub, and the kimi-k3 price dropping on OpenRouter to $0.4255/$10.00 per million tokens (cache reads listed at the input price) with kimi-k2.7-code at $0.67/$3.35, and refreshed repository counters. - 2026-10-02 - Recorded the kimi-k3 price rising again on OpenRouter to $2.70/$13.50 per million tokens (cache reads $0.27) hours after the drop, with kimi-k2.7-code unchanged, and refreshed repository counters.
- 2026-10-04 - Recorded the kimi-k3 price dropping again on OpenRouter to $0.72/$13.00 per million tokens (cache reads $0.70), with kimi-k2.7-code unchanged, and refreshed repository counters.
See also #
- Model Selection for Coding Tasks - the Kimi pricing this note’s economics ride on
- Agentic Coding Tools Landscape - where a vendor-built CLI sits in the harness layer
- Codex - the subscription-included big-vendor CLI comparison point
- Protocols Feature Matrix - ACP and MCP, both of which Kimi Code speaks
References #
https://github.com/MoonshotAI/kimi-code - the canonical repository: license, features, subagents, ACP
https://moonshotai.github.io/kimi-code/en/ - official documentation
https://frontierharness.org - the independent pass-rate and cost-per-task numbers
https://hn.algolia.com/api/v1/items/45767884 - the largest HN thread, cited as the thin-footprint signal
https://openrouter.ai/api/v1/models - live per-token prices grounding the kimi-k3 and kimi-k2.7-code figures, as of 2026-10-04