Skip to main content

​
Structured Output

Using the response_format parameter, you can have the model return structured data according to a specified JSON Schema, ensuring reliable and parseable output format.

​
JSON Mode

The simplest form of structured output, forcing the model to return valid JSON: 
curl https://crazyrouter.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role": "system", "content": "You are a data extraction assistant. Return results in JSON format."},
      {"role": "user", "content": "Extract the names and job titles from the following text: John is the CTO of the company, and Jane is the product manager."}
    ],
    "response_format": {"type": "json_object"}
  }'

​
Response

{
  "people": [
    {"name": "John", "title": "CTO"},
    {"name": "Jane", "title": "Product Manager"}
  ]
}

​
JSON Schema Mode

Using the json_schema type, you can precisely define the output structure, and the model will strictly follow the schema: 
curl https://crazyrouter.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role": "user", "content": "Analyze the quality of this code: def add(a, b): return a + b"}
    ],
    "response_format": {
      "type": "json_schema",
      "json_schema": {
        "name": "code_review",
        "strict": true,
        "schema": {
          "type": "object",
          "properties": {
            "score": {
              "type": "integer",
              "description": "Code quality score, 1-10"
            },
            "issues": {
              "type": "array",
              "items": {
                "type": "object",
                "properties": {
                  "severity": {
                    "type": "string",
                    "enum": ["low", "medium", "high"]
                  },
                  "description": {
                    "type": "string"
                  }
                },
                "required": ["severity", "description"],
                "additionalProperties": false
              }
            },
            "suggestions": {
              "type": "array",
              "items": {"type": "string"}
            }
          },
          "required": ["score", "issues", "suggestions"],
          "additionalProperties": false
        }
      }
    }
  }'

​
Response

{
  "score": 7,
  "issues": [
    {
      "severity": "low",
      "description": "Missing type annotations"
    },
    {
      "severity": "low",
      "description": "Missing docstring"
    }
  ],
  "suggestions": [
    "Add type annotations: def add(a: int, b: int) -> int",
    "Add a docstring describing the function's purpose",
    "Consider adding input validation"
  ]
}
When using json_schema mode, setting "strict": true ensures the model strictly follows the schema definition. additionalProperties must be set to false.
When using json_object mode, you must explicitly prompt the model to return JSON in the system or user message, otherwise it may result in infinite generation.