Structured outputs

Request a predictable JSON shape for results your application will process.

Choose a schema-capable route

Use response_format.type: json_schema with a named json_schema object. A model must publish structured_outputs support. JSON object mode only requests JSON; it does not enforce your application's field schema.

Define a small output shape

The example requires a single string field and rejects extra properties. Keep the schema within the provider's supported JSON Schema subset and use provider.require_parameters: true when the format is essential.

curl

curl --request POST \
  --url 'https://api.tokenlane.de/v1/chat/completions' \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
  "model": "deepseek/deepseek-v4-flash-0731",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ],
  "max_tokens": 400,
  "response_format": {
    "type": "json_schema",
    "json_schema": {
      "name": "answer",
      "strict": true,
      "schema": {
        "type": "object",
        "properties": {
          "answer": {
            "type": "string"
          }
        },
        "required": [
          "answer"
        ],
        "additionalProperties": false
      }
    }
  },
  "provider": {
    "require_parameters": true
  }
}'

node

import OpenAI from 'openai';

const client = new OpenAI({ baseURL: "https://api.tokenlane.de/v1", apiKey: process.env.TOKENLANE_API_KEY });
const response = await client.chat.completions.create({
  "model": "deepseek/deepseek-v4-flash-0731",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ],
  "max_tokens": 400,
  "response_format": {
    "type": "json_schema",
    "json_schema": {
      "name": "answer",
      "strict": true,
      "schema": {
        "type": "object",
        "properties": {
          "answer": {
            "type": "string"
          }
        },
        "required": [
          "answer"
        ],
        "additionalProperties": false
      }
    }
  },
  "provider": {
    "require_parameters": true
  }
}, {});
console.log(response);

python

import os
import json
from openai import OpenAI

client = OpenAI(base_url="https://api.tokenlane.de/v1", api_key=os.environ["TOKENLANE_API_KEY"])
body = json.loads(r'''{
  "model": "deepseek/deepseek-v4-flash-0731",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ],
  "max_tokens": 400,
  "response_format": {
    "type": "json_schema",
    "json_schema": {
      "name": "answer",
      "strict": true,
      "schema": {
        "type": "object",
        "properties": {
          "answer": {
            "type": "string"
          }
        },
        "required": [
          "answer"
        ],
        "additionalProperties": false
      }
    }
  },
  "provider": {
    "require_parameters": true
  }
}''')
response = client.chat.completions.create(
    model=body["model"],
    messages=body["messages"],
    max_tokens=body["max_tokens"],
    response_format=body["response_format"],
    extra_body=json.loads(r'''{
  "provider": {
    "require_parameters": true
  }
}''')
)
print(response)

vercel

// This operation uses HTTP directly.
const response = await fetch("https://api.tokenlane.de/v1/chat/completions", {
  method: "POST",
  headers: {
    "Authorization": "Bearer " + process.env.TOKENLANE_API_KEY,
    "Content-Type": "application/json",
  },
  body: `{
  "model": "deepseek/deepseek-v4-flash-0731",
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ],
  "max_tokens": 400,
  "response_format": {
    "type": "json_schema",
    "json_schema": {
      "name": "answer",
      "strict": true,
      "schema": {
        "type": "object",
        "properties": {
          "answer": {
            "type": "string"
          }
        },
        "required": [
          "answer"
        ],
        "additionalProperties": false
      }
    }
  },
  "provider": {
    "require_parameters": true
  }
}`,
});
console.log(await response.text());

Validate the answer

Parse the content and validate it against your schema before using it. Check truncation, refusal and errors first. A schema-valid string is still untrusted content and can be factually wrong. Lower output limits can cut off JSON; leave room for the complete object.

Fix format failures

Simplify unsupported schema constructs or choose a compatible route. JSON healing can repair syntax on non-streaming requests, but it does not validate your schema or make incorrect content correct. Do not retry indefinitely after the same unsupported-schema error.