Send chat completion requests to any supported LLM model through an MPP session.

Start an MPP session against https://openrouter.mpp.tempo.xyz/v1/chat/completions, then POST JSON body. The model field determines which provider and model to use.

{
  "body": {
    "model": "string (e.g. 'openai/gpt-4', 'anthropic/claude-3-opus', 'meta-llama/llama-3-70b')",
    "messages": [
      {
        "role": "string ('system'|'user'|'assistant')",
        "content": "string"
      }
    ],
    "temperature": "number (0-2, optional)",
    "max_tokens": "number (optional)",
    "top_p": "number (optional)",
    "stream": "boolean (optional)",
    "response_format": "object (optional)",
    "tools": "array (optional, function calling)"
  }
}

Send text completion requests.

POST JSON body. Legacy text completion API. Use /chat/completions for chat-based models.

{
  "body": {
    "model": "string",
    "prompt": "string",
    "max_tokens": "number (optional)",
    "temperature": "number (optional)"
  }
}

Generate embeddings for text input.

POST JSON body. Generate vector embeddings for text. Supports OpenAI-compatible embedding models.

{
  "body": {
    "model": "string (e.g. 'openai/text-embedding-3-small')",
    "input": "string | array of strings"
  }
}

Send messages (Anthropic-compatible).

POST JSON body. Anthropic Messages API compatible endpoint. Works like /chat/completions but uses Anthropic message format.

{
  "body": {
    "model": "string",
    "max_tokens": "number",
    "messages": [
      {
        "role": "string",
        "content": "string"
      }
    ],
    "temperature": "number (optional)",
    "stream": "boolean (optional)"
  }
}

Send a response request (OpenAI Responses API compatible).

POST JSON body. OpenAI Responses API compatible. Send input for model response generation.

{
  "body": {
    "model": "string",
    "input": "string | array",
    "instructions": "string (optional)",
    "temperature": "number (optional)",
    "max_output_tokens": "number (optional)"
  }
}