Kimi API: Setup, API Keys & First Request

1. 1

   Get a key

   Sign in at platform.kimi.ai, add at least $1 and create an API key.

2. 2

   Point your code at Kimi

   Use the OpenAI SDK with Kimi's base URL and your key.

3. 3

   Send a message

   Call chat completions with the model kimi-k3.

Base URL: https://api.moonshot.ai/v1

Not a developer? You probably want the Kimi app, not the API. See plan or API: how to choose and how to use Kimi AI.

1. Go to platform.kimi.ai. Check the address before you sign in.
2. Sign up or sign in.
3. You land in the console. This is where you manage keys, billing and usage.

Open the billing page in the console and top up. The minimum is $1, which is plenty for testing.

On a small screen, swipe the table sideways to see every column.

• New accounts are slow on purpose: 3 requests a minute. Limits rise as your total top-ups grow.
• Once your top-ups reach $5, Moonshot adds a $5 voucher.
• Kimi K3 needs at least a $1 top-up before it will answer.

1. In the console, open API Keys.
2. Create a new key. Give it a name you will recognise, such as "laptop-test".
3. Copy the key straight away and keep it somewhere safe, such as a password manager.

Do not paste the key into your code. Save it as an environment variable named MOONSHOT_API_KEY, which is the name Moonshot's own examples use. Your code then reads it from there.

Mac and Linux

export MOONSHOT_API_KEY="YOUR_KIMI_API_KEY"

Windows PowerShell

setx MOONSHOT_API_KEY "YOUR_KIMI_API_KEY"

Replace YOUR_KIMI_API_KEY with your real key. On Windows, close PowerShell and open it again so the new value is picked up. On Mac and Linux, the export line lasts until you close the terminal.

Pick the way you prefer. All three do the same thing: ask Kimi K3 "What is 1+1?" and print the answer.

curl https://api.moonshot.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $MOONSHOT_API_KEY" \
  -d '{
    "model": "kimi-k3",
    "reasoning_effort": "low",
    "max_completion_tokens": 2048,
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "What is 1+1?"}
    ]
  }'

pip install --upgrade 'openai>=1.0'

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["MOONSHOT_API_KEY"],
    base_url="https://api.moonshot.ai/v1",
)

completion = client.chat.completions.create(
    model="kimi-k3",
    reasoning_effort="low",
    max_completion_tokens=2048,
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is 1+1?"},
    ],
)

print(completion.choices[0].message.content)

npm install openai@latest

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.MOONSHOT_API_KEY,
  baseURL: "https://api.moonshot.ai/v1",
});

const completion = await client.chat.completions.create({
  model: "kimi-k3",
  reasoning_effort: "low",
  max_completion_tokens: 2048,
  messages: [
    { role: "system", content: "You are a helpful assistant." },
    { role: "user", content: "What is 1+1?" },
  ],
});

console.log(completion.choices[0].message.content);

{
  "id": "chatcmpl-...",
  "model": "kimi-k3",
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "1 + 1 = 2",
        "reasoning_content": "..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": ...,
    "completion_tokens": ...,
    "total_tokens": ...
  }
}

For multi-step conversations with Kimi K3, send back the complete assistant message each turn, including its reasoning_content, not just the text.

To switch model, change the model value. Prices are per million tokens, in US dollars.

On a small screen, swipe the table sideways to see every column.

Older names such as kimi-k2.5 and moonshot-v1 were retired on 31 August 2026 and now return a 404 error. Settings differ between models, so check the docs when you switch. Guides: Kimi K3 API · all Kimi models · API docs guide

Almost nothing. The test above sends a few dozen tokens and caps the answer at 2,048 tokens. Even if Kimi K3 used the whole cap, that is about 3 cents. A normal "1+1" answer costs a tiny part of that.

Kimi API pricing · Web Search API pricing · Enterprise pricing

1. 1

   Never put it in code

   Read it from an environment variable or a secrets manager.

2. 2

   Never commit it

   Keep keys out of Git. Add .env files to .gitignore.

3. 3

   Never ship it to browsers

   Call Kimi from your server, not from web page or app code that users can read.

4. 4

   One key per project

   If one leaks, you can delete it without breaking the others.

5. 5

   Watch your usage

   Check the console now and then for spending you do not recognise.

6. 6

   Leaked? Replace it

   Delete the key in the console and create a new one straight away.

On a small screen, swipe the table sideways to see every column.

More fixes: Kimi AI not working.

• Create an account at platform.kimi.ai
• Top up at least $1
• Create an API key and store it safely
• Set the MOONSHOT_API_KEY environment variable
• Install the OpenAI SDK, or use curl
• Send the first request and see an answer
• Set a token cap in every request

Your ticks are not saved if you close or reload this page.