What the Kimi K3 API costs. And Kimi K3, K2.7 Code and K2.6 for 10 to 21% less.

Moonshot's prices per million tokens for Kimi K3, K2.7 Code and K2.6, next to what APIsRouter charges for the ones it sells. Call them through the OpenAI SDK with one base URL and one key.

The same traffic for 30 days

Tokens per request: agent 68,000 in (60,000 cached) and 3,000 out; code review 23,000 in (20,000 cached) and 1,000 out; chatbot 3,000 in (2,000 cached) and 500 out.

MoonshotAPIsRouter
Research agent, 200 tasks a day (Kimi K3)$522$453
Code review bot, 1,000 reviews a day (Kimi K2.7 Code)$320$272
Support chatbot, 1,000 answers a day (Kimi K2.6)$98$86
Total$940$811

Kimi API prices, per million tokens

Moonshot's list for the Kimi models. Where APIsRouter sells a model below Moonshot, both prices are shown. Moonshot prices from platform.kimi.ai/docs/pricing/chat, September 2026, before tax.

ModelInput: Moonshot / APIsRouterCache hit: Moonshot / APIsRouterOutput: Moonshot / APIsRouter
Kimi K3$3 / $2.50$0.30 / $0.25$15 / $13.50
Kimi K2.7 Code$0.95 / $0.85$0.19 / $0.15$4 / $3.50
Kimi K2.6$0.95 / $0.85$0.16 / $0.14$4 / $3.50
Kimi K2.7 Code Highspeed$1.90 / Not on APIsRouter$0.38 / Not on APIsRouter$8 / Not on APIsRouter

How Kimi API billing works

  • Cache hits are automatic: Kimi caches repeated prompt prefixes without any setting, and cached tokens bill at the cache-hit price: $0.30 instead of $3 per million on Kimi K3 at Moonshot, $0.25 on APIsRouter.
  • Kimi K2.7 Code for coding: Kimi K2.7 Code is Moonshot's coding model, at $0.85 / $3.50 per million input and output tokens on APIsRouter against $2.50 / $13.50 for Kimi K3. On APIsRouter the Kimi models take OpenAI-format requests, so they work in OpenAI-compatible tools, not in Claude Code.
  • One key for Kimi and the rest: The same APIsRouter key calls Claude, GPT, Gemini, DeepSeek, GLM and Qwen, all paid from one balance, with no Moonshot account.
  • Pay as you go: No subscription needed: top up from $10, with +30% on the first top-up. Unused credit never expires.

Switch with one line

  • Get a key: Start free on the $2 trial credit, or top up from $10 below. A top-up key arrives by email.
  • Point the OpenAI SDK at APIsRouter: Set base_url to https://api.apisrouter.com/v1. Model IDs are Moonshot's, such as kimi-k3.
  • Send requests: Every request shows up in your console with its tokens and cost.
from openai import OpenAI

client = OpenAI(
    base_url="https://api.apisrouter.com/v1",
    api_key="sk-your-apisrouter-key",
)

reply = client.chat.completions.create(
    model="kimi-k3",
    messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)

Two ways to pay

Both fill one balance, spent at the prices above.

FAQ

How much does the Kimi K3 API cost?

At Moonshot, $3 / $15 per million input and output tokens, and $0.30 for cache hits. On APIsRouter, $2.50 / $13.50 and $0.25.

How do I get a Kimi API key?

Sign up at platform.kimi.ai for a Moonshot key, or use one APIsRouter key for Kimi and every other model: start on the $2 trial credit from your dashboard, or top up from $10 and get the key by email.

Is the Kimi API free?

Moonshot bills the API per token. On APIsRouter you can start on a $2 trial credit: sign in and claim it from your dashboard, no card needed.

Which Kimi model should I use?

Kimi K3 is Moonshot's newest Kimi model and the priciest per token. Kimi K2.7 Code is its coding model, and Kimi K2.6 the one before it; both cost $0.85 / $3.50 per million input and output tokens on APIsRouter.

Is the Kimi API OpenAI-compatible?

Yes. Set base_url (baseURL in Node) to https://api.apisrouter.com/v1, use your APIsRouter key and Moonshot's model IDs, such as kimi-k3. On APIsRouter the Kimi models take OpenAI-format requests only, so use them from OpenAI-compatible tools rather than Claude Code.

Do cached tokens cost less?

Yes. Kimi caches repeated prompt prefixes on its own, and cached input bills at $0.25 instead of $2.50 per million on Kimi K3 through APIsRouter.