Structured output

response_format on /v1/chat/completions asks the model for JSON:

`response_format`	Reply
`{"type": "json_schema", "json_schema": {...}}`	JSON that follows your schema
`{"type": "json_object"}`	Some valid JSON object (JSON mode); describe the shape you want in the prompt

Models: the models marked “json” on the models page support it (every chat model except meta-llama/llama-4-maverick). On other models the field may be ignored, so validate the reply.

JSON Schema

curl https://api.thetoken.dev/v1/chat/completions \
  -H "Authorization: Bearer $TOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-oss-120b",
    "messages": [{"role": "user", "content": "Extract: Sara, 34, lives in Jeddah."}],
    "response_format": {
      "type": "json_schema",
      "json_schema": {
        "name": "person",
        "strict": true,
        "schema": {
          "type": "object",
          "properties": {
            "name": {"type": "string"},
            "age": {"type": "integer"},
            "city": {"type": "string"}
          },
          "required": ["name", "age", "city"],
          "additionalProperties": false
        }
      }
    }
  }'

import json
import os
from openai import OpenAI

client = OpenAI(base_url="https://api.thetoken.dev/v1", api_key=os.environ["TOKEN_API_KEY"])

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

response = client.chat.completions.create(
    model="openai/gpt-oss-120b",
    messages=[{"role": "user", "content": "Extract: Sara, 34, lives in Jeddah."}],
    response_format={"type": "json_schema", "json_schema": {"name": "person", "strict": True, "schema": schema}},
)
person = json.loads(response.choices[0].message.content)
print(person["name"], person["age"], person["city"])

import OpenAI from 'openai'

const client = new OpenAI({ baseURL: 'https://api.thetoken.dev/v1', apiKey: process.env.TOKEN_API_KEY })

const response = await client.chat.completions.create({
  model: 'openai/gpt-oss-120b',
  messages: [{ role: 'user', content: 'Extract: Sara, 34, lives in Jeddah.' }],
  response_format: {
    type: 'json_schema',
    json_schema: {
      name: 'person',
      strict: true,
      schema: {
        type: 'object',
        properties: { name: { type: 'string' }, age: { type: 'integer' }, city: { type: 'string' } },
        required: ['name', 'age', 'city'],
        additionalProperties: false,
      },
    },
  },
})
const person = JSON.parse(response.choices[0].message.content ?? '{}')
console.log(person)

The reply’s content is a JSON string:

{"name": "Sara", "age": 34, "city": "Jeddah"}

JSON mode

curl https://api.thetoken.dev/v1/chat/completions \
  -H "Authorization: Bearer $TOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4.1-flash",
    "messages": [
      {"role": "system", "content": "Reply with a JSON object with keys \"sentiment\" and \"score\"."},
      {"role": "user", "content": "The delivery was late but the support team was great."}
    ],
    "response_format": {"type": "json_object"}
  }'

Good to know

A reply cut off by max_tokens (finish_reason: "length") is incomplete JSON. Leave room for reasoning models, which think before they answer.
On the Responses API, the same option is text.format ({"type": "json_schema", ...}), as in OpenAI’s API.