Switch your client in three lines

Prerequisites: API Key

To use this uncensored api, you need an API key from kimiapi.cc. Visit the Get API key page and sign in with Google or an email and password. The key is displayed immediately and is the only credential required for authentication.

Each account holds one active key; generating a new one replaces the previous key. Your key grants access to the base URL and defines your rate limits. Keep this key secure, as it is tied directly to your prepaid credit balance.

Basic Chat Completion

The core endpoint is POST /v1/chat/completions. Point your client to the base URL and provide the model ID uncensored. This returns a standard text response. You can adjust parameters like temperature or top_p to control creativity.

Copycurl https://api.kimiapi.cc/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "uncensored",
    "messages": [{"role": "user", "content": "Write a blunt product review of a cheap VPN."}]
  }'

Ensure your request body includes the messages array with your prompt. The server processes the input and returns the generated text in the content field of the first choice.

Python SDK

Use the official OpenAI Python SDK to interact with the endpoint. Set the base_url to our API address and provide your kimi api key in the api_key field. This allows you to use familiar Python methods for chat completions.

Copyfrom openai import OpenAI

client = OpenAI(base_url="https://api.kimiapi.cc/v1", api_key="YOUR_KEY")

resp = client.chat.completions.create(
    model="uncensored",
    messages=[{"role": "user", "content": "Summarise this thread without softening it."}],
)
print(resp.choices[0].message.content)

Instantiate the client with your credentials and call chat.completions.create(). The response object provides access to the generated content, token usage, and other metadata. This approach ensures you are using the same client code you would use for other OpenAI-compatible services.

Node SDK

For JavaScript developers, the @anthropic-ai/sdk or official openai Node package works with minor configuration. Configure the baseURL to point to our endpoint and set the apiKey to your key. This enables seamless integration into Node.js applications.

Copyimport OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://api.kimiapi.cc/v1", apiKey: process.env.API_KEY });

const resp = await client.chat.completions.create({
  model: "uncensored",
  messages: [{ role: "user", content: "Draft a villain monologue for my game." }],
});
console.log(resp.choices[0].message.content);

Create a completion request by passing the model ID uncensored and your messages. The SDK handles the HTTP requests and parses the JSON response. This method is ideal for server-side rendering or API integrations where you need programmatic access to the model.

Streaming Responses

Enable streaming by setting stream: true in your request. The API returns a Server-Sent Events (SSE) stream, delivering tokens as they are generated. This reduces perceived latency for interactive applications.

Copystream = client.chat.completions.create(
    model="uncensored",
    messages=[{"role": "user", "content": "Tell the story in second person."}],
    stream=True,
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

Each chunk contains partial content. The final chunk includes the complete token usage statistics. Handle the stream events to accumulate the full response. This feature is essential for chat interfaces where users expect to see text appear in real time.

Limits, Errors & Context

Each key is limited to 300 requests per minute and 8 concurrent requests. The request body must not exceed 8 MB. If your key is invalid, you receive a 401 error. A 402 error indicates insufficient prepaid credit. A 429 error signals that you have exceeded the rate limit.

The context window is 64,000 tokens total, including prompt and completion. If you do not specify max_tokens, the output is capped at 2,048 tokens. Adjust these limits based on your application's needs to ensure reliable performance.