Kimi Code plans: prices, limits and API access

Kimi Code plans and prices

Kimi Code has no separate price of its own and comes included with every Kimi membership tier from Plus up, while the Free and Go tiers have no coding quota. Switching to yearly billing lowers the monthly price of all four paid tiers and saves up to $480 a year on Ultra ($1,908 billed yearly compared to $2,388 paid month by month).

How Kimi Code usage limits work

• New Kimi Code plans don't have a weekly quota limit and instead run on a rolling 5-hour rate window, so sending too many requests in a short burst triggers rate limiting that lifts on its own once the 5-hour window rolls over.
• Your Kimi membership's monthly total quota still applies on top of the 5-hour window, and if you hit that monthly ceiling, Kimi Code is frozen until the monthly quota resets or you upgrade your plan.
• Kimi Code shares one common quota pool with the rest of your Kimi membership, so requests from the Kimi Code CLI, VS Code, Kimi Code Desktop, third-party coding tools, every signed-in device and every Kimi Code API key all draw from the same balance.
• Kimi describes each tier's coding quota in words rather than publishing token or message counts, going from "regularly refreshed usage quotas" on Plus and "ample limits and increased concurrency caps" on Pro to "generous credits" on Max and "our highest credit allowance" on Ultra, while the membership page lists agent credits at 2x on Pro, 5x on Max and 10x on Ultra.

Models on each Kimi Code plan

Kimi Code exposes six model ids, and your membership tier decides both which models you can call and whether Kimi K3 and K3.1 Pro get the full 1M-token context window.

K3.1 Pro rolled out in Kimi Code on October 23, 2026, runs in thinking mode only with reasoning_effort set to low, high or max (high by default), and Kimi warns on its pricing page that K3.1 Pro requests may queue or fail at launch. The Highspeed variants and the 1M-context k3 model burn through your quota faster than standard models: k3.1-pro-highspeed uses about 2x quota, kimi-for-coding-highspeed uses 3x quota, and k3 on its 1M window uses about twice the quota of k3-256k.

Extra Usage when your quota runs out

When your subscription quota runs out, you can turn on Extra Usage, which is a pay-as-you-go balance denominated in yuan (RMB) and shared between Kimi on the web and Kimi Code. Extra Usage is priced close to the Kimi Open Platform API, gets deducted only after your time-limited subscription quota is spent, and lets you set an optional monthly spending cap so a busy agent loop doesn't drain your balance.

Top-ups start at 25 yuan, with limits of up to 10 top-ups a day, 3,000 yuan a day, and a maximum balance cap of 10,000 yuan, and the balance never expires and is generally non-refundable. You can check your remaining quota, rate-limit status, API keys and devices in the Kimi Code Console at kimi.com/code/console, or run /usage inside the Kimi Code CLI.

Use Kimi Code in other coding tools

If you are comparing a Kimi coding plan with other AI coding plans to run in third-party agents like Claude Code, OpenCode, Codex or Hermes Agent, you can hook your subscription straight into those tools with a Kimi Code API key.

1. Open the Kimi Code Console at kimi.com/code/console and generate a new Kimi Code API key.
2. Point your coding tool at the overseas OpenAI-compatible base URL https://api.kimi.ai/coding/v1 (which supports both Responses and Chat Completions) or the Anthropic-compatible base URL https://api.kimi.ai/coding/, swapping api.kimi.com in place of api.kimi.ai if you are connecting from China.
3. Configure the tool with an exact model id from the table above, such as k3.1-pro or k3, rather than the model's display name.

Use your Kimi plan as an OpenAI-compatible API

Plan2API connects your Kimi membership when you sign up at plan2api.com/connect and paste a Kimi Code API key, turning your subscription into a standard OpenAI-compatible Kimi API at https://api.plan2api.com/v1 (POST /v1/chat/completions and GET /v1/models). It serves Kimi K3 under the model id kimi/k3 (or the alias moonshotai/kimi-k3) with a 1,048,576-token context window, text and image input, and low, high or max reasoning effort, starting conversations on the 256K window and automatically shifting to the 1M window once a thread passes about 192K tokens, with all requests drawing from your same Kimi Code quota.

Plan2API's Kimi catalog only lists kimi/k3, so K3.1 Pro and the Highspeed models aren't available through it, though the same p2a_ key (sent in an Authorization: Bearer header) can also route requests to connected Claude, ChatGPT and Gemini subscriptions. The Free plan costs $0 with no card required and covers 1 connected subscription and 50 chat requests a day, while Pro is $12 a month (or $99 a year) for 3 connected subscriptions and no daily request cap. A curl example follows below.

curl https://api.plan2api.com/v1/chat/completions \
  -H "Authorization: Bearer $PLAN2API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi/k3",
    "reasoning_effort": "high",
    "messages": [
      {"role": "user", "content": "Find the bug in this function and explain the fix."}
    ]
  }'