JSON Mode

Guarantee valid, structured JSON output from AI models. Perfect for data extraction, automation, and API integrations.

❌ Without JSON Mode

Here's the extracted data:

Name: John Smith
Email: john@example.com
Company: Acme Corp

Let me know if you need anything else!

✅ With JSON Mode

{
  "name": "John Smith",
  "email": "john@example.com",
  "company": "Acme Corp"
}

Two Ways to Get JSON

RECOMMENDED

Structured Outputs

Define a JSON Schema and get guaranteed schema-compliant output. Best for production use with strict requirements.

BASIC

JSON Mode

Guarantees valid JSON but without schema validation. Simpler to use when you just need parseable JSON.

Structured Outputs (Recommended)

Define exactly what shape you want the output to be:

Python

                    from pydantic import BaseModel

# Define your schema with Pydantic
class ContactInfo(BaseModel):
    name: str
    email: str
    company: str | None

# Request structured output
response = client.beta.chat.completions.parse(
    model="mythic-4",
    messages=[
        {"role": "user", "content": "Extract: John Smith from Acme Corp, john@example.com"}
    ],
    response_format=ContactInfo
)

# Access parsed data directly
contact = response.choices[0].message.parsed
print(contact.name)     # "John Smith"
print(contact.email)    # "john@example.com"
print(contact.company)  # "Acme Corp"
                

💡 Type Safety

With Pydantic models, you get full IDE autocomplete and type checking. The response is automatically parsed into your model class.

JSON Mode (Basic)

For simpler use cases where you just need valid JSON:

Python

                    response = client.chat.completions.create(
    model="mythic-4",
    messages=[
        {"role": "system", "content": "Output valid JSON"},
        {"role": "user", "content": "List 3 programming languages with their use cases"}
    ],
    response_format={"type": "json_object"}
)

# Parse the JSON response
import json
data = json.loads(response.choices[0].message.content)
print(data)
                

⚠️ Important

When using JSON mode, you must instruct the model to produce JSON in your prompt (system or user message). Otherwise, the model may generate endless whitespace.

Comparison

Feature	Structured Outputs	JSON Mode
Valid JSON	✓ Always	✓ Always
Schema validation	✓ Guaranteed	Not guaranteed
Required keys	✓ Enforced	Not enforced
Type safety	✓ With Pydantic	Manual parsing
Setup	Define schema	Just enable
Best for	Production APIs	Quick prototypes

Use Cases

📧

Email Extraction

Parse names, emails, and details from text

🏷️

Classification

Categorize content with structured labels

📊

Data Pipelines

Transform unstructured to structured data

🔌

API Responses

Generate consistent API payloads

📝

Form Generation

Create structured form data from descriptions

🧪

Test Data

Generate mock data matching your schema

JSON Schema with response_format

You can also use JSON Schema directly:

Python

                    response = client.chat.completions.create(
    model="mythic-4",
    messages=[
        {"role": "user", "content": "Generate a product with name, price, and category"}
    ],
    response_format={
        "type": "json_schema",
        "json_schema": {
            "name": "product",
            "strict": True,
            "schema": {
                "type": "object",
                "properties": {
                    "name": {"type": "string"},
                    "price": {"type": "number"},
                    "category": {"type": "string"}
                },
                "required": ["name", "price", "category"],
                "additionalProperties": False
            }
        }
    }
)
                