Kimi K3

Kimi K3 is built on a sparse Mixture-of-Experts architecture that activates 104B of its 2.8 trillion parameters per token — routing 16 of 896 experts — keeping latency and cost low while preserving deep reasoning ability. Configurable low, medium, and high reasoning efforts let you trade speed for depth on a per-request basis.

The model supports function calling, structured outputs, and image input, and streams responses over an OpenAI-compatible API. Prompt prefixes are cached automatically — repeated system prompts and long documents are billed at the cache-read rate with no extra configuration.

import OpenAI from "openai";

const openai = new OpenAI({
  baseURL: "https://api.inference.net/v1",
  apiKey: "<YOUR_API_KEY>",
});

const completion = await openai.chat.completions.create({
  model: "kimi-k3",
  messages: [
    {
      role: "user",
      content: "What is the meaning of life?"
    }
  ],
  stream: true,
});

for await (const chunk of completion) {
  process.stdout.write(chunk.choices[0]?.delta.content as string);
}

import OpenAI from "openai";

const openai = new OpenAI({
  baseURL: "https://api.inference.net/v1",
  apiKey: "<YOUR_API_KEY>",
});

const completion = await openai.chat.completions.create({
  model: "kimi-k3",
  messages: [
    {
      role: "user",
      content: "What is the meaning of life?"
    }
  ],
  stream: true,
});

for await (const chunk of completion) {
  process.stdout.write(chunk.choices[0]?.delta.content as string);
}