> ## Documentation Index
> Fetch the complete documentation index at: https://docs.oxen.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Kimi K3

> Agentic coding and vision, 1M context

<CardGroup cols={1}>
  <Card title="Try Kimi K3 in the Workbench" icon="flask" href="https://www.oxen.ai/ai/workbench?model=kimi-k3">
    Run this model interactively, tune parameters, and compare outputs.
  </Card>
</CardGroup>

**Model ID:** `kimi-k3`

Kimi K3 is Moonshot AI's flagship Mixture-of-Experts model with a 1M-token context window, built for long-horizon agentic coding and reasoning workloads with native visual understanding.

The model uses the Kimi Delta Attention (KDA) hybrid linear attention mechanism and activates 16 of 896 experts per token. Thinking mode is always enabled: responses include separate reasoning content alongside the answer, and the `reasoning_effort` parameter currently accepts only "max". It handles tool calling (including forced tool choice and strict JSON schema output) and accepts images and videos through base64 or the Files API. Sampling parameters are fixed by the provider (temperature 1.0, top\_p 0.95), and input caching is priced separately, with cache hits costing a tenth of cache misses. Open weights are slated for release in late July 2026.

| Metric             | Value                          |
| ------------------ | ------------------------------ |
| Parameter Count    | 2.8 trillion                   |
| Mixture of Experts | Yes (16 of 896 experts active) |
| Context Length     | 1,048,576 tokens               |
| Multilingual       | Yes                            |
| Quantized\*        | Unknown                        |

\**Quantization is specific to the inference provider and the model may be offered with different quantization levels by other providers.*

## Example request

<Tip>
  Use the [Workbench](https://www.oxen.ai/ai/workbench?model=kimi-k3) as a request builder: configure parameters for this model in the UI, then open the **API** tab to copy the exact cURL or Python call.
</Tip>

<Tabs>
  <Tab title="Minimal">
    <CodeGroup>
      ```bash cURL theme={null}
      curl -X POST https://hub.oxen.ai/api/ai/chat/completions \
        -H "Content-Type: application/json" \
        -H "Authorization: Bearer $OXEN_API_KEY" \
        -d '{
        "model": "kimi-k3",
        "messages": [
          {
            "role": "user",
            "content": "Hello, what can you do?"
          }
        ]
      }'
      ```

      ```python Python theme={null}
      import os
      import requests

      response = requests.post(
          "https://hub.oxen.ai/api/ai/chat/completions",
          headers={
              "Content-Type": "application/json",
              "Authorization": f"Bearer {os.environ['OXEN_API_KEY']}",
          },
          json={
              "model": "kimi-k3",
              "messages": [
                  {
                      "role": "user",
                      "content": "Hello, what can you do?"
                  }
              ]
          },
      )
      response.raise_for_status()
      print(response.json())
      ```
    </CodeGroup>
  </Tab>

  <Tab title="Basic parameters">
    <CodeGroup>
      ```bash cURL theme={null}
      curl -X POST https://hub.oxen.ai/api/ai/chat/completions \
        -H "Content-Type: application/json" \
        -H "Authorization: Bearer $OXEN_API_KEY" \
        -d '{
        "model": "kimi-k3",
        "messages": [
          {
            "role": "user",
            "content": "Hello, what can you do?"
          }
        ],
        "temperature": 0.7,
        "max_tokens": 1024,
        "stream": false
      }'
      ```

      ```python Python theme={null}
      import os
      import requests

      response = requests.post(
          "https://hub.oxen.ai/api/ai/chat/completions",
          headers={
              "Content-Type": "application/json",
              "Authorization": f"Bearer {os.environ['OXEN_API_KEY']}",
          },
          json={
              "model": "kimi-k3",
              "messages": [
                  {
                      "role": "user",
                      "content": "Hello, what can you do?"
                  }
              ],
              "temperature": 0.7,
              "max_tokens": 1024,
              "stream": false
          },
      )
      response.raise_for_status()
      print(response.json())
      ```
    </CodeGroup>
  </Tab>

  <Tab title="All parameters">
    <CodeGroup>
      ```bash cURL theme={null}
      curl -X POST https://hub.oxen.ai/api/ai/chat/completions \
        -H "Content-Type: application/json" \
        -H "Authorization: Bearer $OXEN_API_KEY" \
        -d '{
        "model": "kimi-k3",
        "messages": [
          {
            "role": "user",
            "content": "Hello, what can you do?"
          }
        ],
        "temperature": 0.7,
        "max_tokens": 1024,
        "stream": false,
        "top_p": 1.0
      }'
      ```

      ```python Python theme={null}
      import os
      import requests

      response = requests.post(
          "https://hub.oxen.ai/api/ai/chat/completions",
          headers={
              "Content-Type": "application/json",
              "Authorization": f"Bearer {os.environ['OXEN_API_KEY']}",
          },
          json={
              "model": "kimi-k3",
              "messages": [
                  {
                      "role": "user",
                      "content": "Hello, what can you do?"
                  }
              ],
              "temperature": 0.7,
              "max_tokens": 1024,
              "stream": false,
              "top_p": 1.0
          },
      )
      response.raise_for_status()
      print(response.json())
      ```
    </CodeGroup>
  </Tab>
</Tabs>

## Fetch model details

The [models endpoint](/inference-api/reference/models/overview) returns the full model object, including its `json_request_schema`.

```bash theme={null}
curl -H "Authorization: Bearer $OXEN_API_KEY" https://hub.oxen.ai/api/ai/models/kimi-k3
```

## Request parameters

This model follows the standard OpenAI chat completions request body. See the [chat completions reference](../inference-api.mdx) for the full parameter list.