Docs/AI SDK

AI

Structured Output

Last updated March 3, 2026

Get type-safe, validated JSON responses from AI models.

Structured output ensures AI responses conform to a specific schema, making it easy to parse and use in your application.

Basic Usage

Codetext
import { z } from 'zod';
 
const ProductSchema = z.object({
  name: z.string(),
  price: z.number(),
  category: z.enum(['electronics', 'clothing', 'food']),
  inStock: z.boolean()
});
 
const response = await cencori.ai.chat({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Generate a product' }],
  responseFormat: {
    type: 'json_schema',
    json_schema: {
      name: 'product',
      schema: zodToJsonSchema(ProductSchema)
    }
  }
});
 
// Response is guaranteed to match schema
const product = ProductSchema.parse(JSON.parse(response.content));
console.log(product.name);   // Type-safe access
console.log(product.price);  // Type: number

JSON Mode

Simple JSON output without schema validation:

Codetext
const response = await cencori.ai.chat({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'List 3 products as JSON' }],
  responseFormat: { type: 'json_object' }
});
 
const data = JSON.parse(response.content);

Schema Examples

Array of Objects

Codetext
const OrdersSchema = z.object({
  orders: z.array(z.object({
    id: z.string(),
    total: z.number(),
    items: z.array(z.object({
      name: z.string(),
      quantity: z.number()
    }))
  }))
});

Nested Structure

Codetext
const ReportSchema = z.object({
  title: z.string(),
  summary: z.string(),
  sections: z.array(z.object({
    heading: z.string(),
    content: z.string(),
    subsections: z.array(z.object({
      heading: z.string(),
      content: z.string()
    })).optional()
  })),
  metadata: z.object({
    wordCount: z.number(),
    readingTime: z.string()
  })
});

Provider Support

ProviderJSON ModeJSON Schema
OpenAI✓✓
Anthropic✓-
Google✓-

Error Handling

Codetext
try {
  const response = await cencori.ai.chat({
    model: 'gpt-4o',
    messages: [{ role: 'user', content: '...' }],
    responseFormat: {
      type: 'json_schema',
      json_schema: { name: 'product', schema: productSchema }
    }
  });
 
  const data = ProductSchema.parse(JSON.parse(response.content));
} catch (error) {
  if (error instanceof z.ZodError) {
    console.error('Schema validation failed:', error.errors);
  }
}