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

# Chat Completions

> Generate chat-based completions using the specified model.

<RequestExample>
  ```bash cURL theme={null}
  curl https://gateway.bud.studio/v1/chat/completions \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "llama-3.2-1b",
      "messages": [
        {
          "role": "user",
          "content": "What is AI?"
        }
      ]
    }'
  ```
</RequestExample>

<ResponseExample>
  ```json 200 theme={null}
  {
    "id": "chatcmpl-abc123",
    "object": "chat.completion",
    "created": 1699000000,
    "model": "llama-3.2-1b",
    "choices": [
      {
        "index": 0,
        "message": {
          "role": "assistant",
          "content": "Hello! How can I help you today?"
        },
        "finish_reason": "stop"
      }
    ],
    "usage": {
      "prompt_tokens": 10,
      "completion_tokens": 20,
      "total_tokens": 30
    }
  }
  ```
</ResponseExample>

## Headers

| Parameter     | Type   | Required | Description                                                                               |
| ------------- | ------ | -------- | ----------------------------------------------------------------------------------------- |
| Authorization | string | Yes      | Bearer authentication header of the form `Bearer <token>` where `<token>` is your API key |

## Body

| Parameter               | Type             | Required | Description                                                                    |
| ----------------------- | ---------------- | -------- | ------------------------------------------------------------------------------ |
| model                   | string           | Yes      | Model identifier (deployment name, adapter name, or routing)                   |
| messages                | array            | Yes      | Array of message objects forming the conversation                              |
| temperature             | float            | No       | Sampling temperature (0.0 to 2.0). Default: model default                      |
| max\_tokens             | integer          | No       | Maximum tokens to generate. Default: model default                             |
| max\_completion\_tokens | integer          | No       | Maximum tokens to generate (alternative to max\_tokens)                        |
| top\_p                  | float            | No       | Nucleus sampling parameter (0.0 to 1.0). Default: 1.0                          |
| frequency\_penalty      | float            | No       | Penalize tokens based on frequency (-2.0 to 2.0). Default: 0.0                 |
| presence\_penalty       | float            | No       | Penalize tokens based on presence (-2.0 to 2.0). Default: 0.0                  |
| repetition\_penalty     | float            | No       | Penalize token repetition (> 0.0). Default: 1.0                                |
| stream                  | boolean          | No       | Enable streaming response. Default: `false`                                    |
| stream\_options         | object           | No       | Streaming options (e.g., `{"include_usage": true}`)                            |
| n                       | integer          | No       | Number of chat completion choices to generate. Default: 1                      |
| stop                    | string or array  | No       | Up to 4 sequences where the API will stop generating                           |
| response\_format        | object           | No       | Response format specification (e.g., `{"type": "json_object"}`)                |
| seed                    | integer          | No       | Random seed for deterministic sampling                                         |
| logprobs                | boolean          | No       | Include log probabilities in response. Default: `false`                        |
| top\_logprobs           | integer          | No       | Number of most likely tokens to return (0-20)                                  |
| logit\_bias             | object           | No       | Modify likelihood of specified tokens appearing                                |
| tools                   | array            | No       | Available tool/function definitions                                            |
| tool\_choice            | string or object | No       | Tool selection strategy. Options: `auto`, `none`, `required`, or specific tool |
| parallel\_tool\_calls   | boolean          | No       | Allow parallel tool calls. Default: `true`                                     |
| user                    | string           | No       | Unique identifier representing your end-user                                   |

### Message Object

```json theme={null}
{
  "role": "system" | "user" | "assistant" | "tool",
  "content": "string or array of content blocks",
  "name": "optional participant name",
  "tool_calls": "array of tool calls (for assistant messages)",
  "tool_call_id": "ID of the tool call (for tool messages)"
}
```

## Supported Providers

<CardGroup cols={4}>
  <Card title="OpenAI" icon="openai">
    GPT-4, GPT-3.5 Turbo, and o1 models
  </Card>

  <Card title="Anthropic" icon="ant">
    Claude 3.5 Sonnet, Claude 3 Opus, and Haiku
  </Card>

  <Card title="Azure OpenAI" icon="microsoft">
    Enterprise GPT-4 and GPT-3.5 deployments
  </Card>

  <Card title="Google" icon="google">
    Gemini Pro and Ultra models
  </Card>

  <Card title="AWS Bedrock" icon="aws">
    Claude, Llama, and Mistral on AWS
  </Card>

  <Card title="Together AI" icon="arrow-progress">
    Llama 3, Mixtral, and open-source models
  </Card>

  <Card title="Fireworks" icon="fire">
    Fast inference for Llama and Mistral models
  </Card>

  <Card title="xAI" icon="x">
    Grok models with extended context
  </Card>
</CardGroup>