Structured Outputs · Docs

Use response_format to get machine-readable output.

JSON Schema

schema = {
    "type": "object",
    "properties": {
        "name": {"type": "string"},
        "context_tokens": {"type": "integer"},
    },
    "required": ["name", "context_tokens"],
    "additionalProperties": False,
}

completion = client.chat.completions.create(
    model="Qwen/Qwen3.8-27B",
    messages=[{"role": "user", "content": "Describe DeepSeek V4 Flash."}],
    response_format={"type": "json_schema", "json_schema": {"name": "model_card", "schema": schema, "strict": True}},
)

JSON mode

{"type": "json_object"} asks for valid JSON without a schema. Tell the model in your prompt what shape you expect.

Notes

Support varies by model — look for Structured outputs in Models.
Always parse and validate the result; if generation hits max_tokens, the JSON can be truncated (finish_reason: "length").