> ## Documentation Index
> Fetch the complete documentation index at: https://www.edenai.co/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Structured Output

> Structured output lets you constrain LLM responses to valid JSON that follows a specific format.

export const TechArticleSchema = ({title, description, path, articleSection, about, proficiencyLevel = "Beginner", dependencies, keywords = [], datePublished, dateModified, image, inLanguage = "en"}) => {
  const baseUrl = "https://www.edenai.co/docs";
  const canonicalUrl = `${baseUrl}/${path}`.replace(/\/+$/, "");
  const ogParams = new URLSearchParams({
    division: articleSection || "",
    title: title || "",
    description: description || ""
  });
  const resolvedImage = image || `https://edenai.mintlify.app/_mintlify/api/og?${ogParams.toString()}`;
  const data = {
    "@context": "https://schema.org",
    "@type": "TechArticle",
    "@id": `${canonicalUrl}#techarticle`,
    mainEntityOfPage: {
      "@type": "WebPage",
      "@id": canonicalUrl
    },
    headline: title,
    name: title,
    description: description,
    url: canonicalUrl,
    inLanguage: inLanguage,
    isPartOf: {
      "@type": "WebSite",
      name: "Eden AI Documentation",
      url: baseUrl
    },
    author: [{
      "@type": "Organization",
      name: "Eden AI",
      url: "https://www.edenai.co/"
    }],
    publisher: {
      "@type": "Organization",
      name: "Eden AI",
      url: "https://www.edenai.co/",
      logo: {
        "@type": "ImageObject",
        url: "https://www.edenai.co/assets/logo.png"
      }
    }
  };
  if (articleSection) data.articleSection = articleSection;
  if (about) data.about = {
    "@type": "Thing",
    name: about
  };
  if (proficiencyLevel) data.proficiencyLevel = proficiencyLevel;
  if (dependencies) data.dependencies = dependencies;
  if (keywords && keywords.length) data.keywords = keywords;
  if (datePublished) data.datePublished = datePublished;
  if (dateModified) data.dateModified = dateModified;
  data.image = Array.isArray(resolvedImage) ? resolvedImage : [resolvedImage];
  const json = JSON.stringify(data);
  const schemaId = `techarticle-${canonicalUrl}`;
  React.useEffect(() => {
    if (typeof document === "undefined") return;
    document.querySelectorAll(`script[data-schema-id="${schemaId}"]`).forEach(n => n.remove());
    const script = document.createElement("script");
    script.type = "application/ld+json";
    script.dataset.schemaId = schemaId;
    script.textContent = json;
    document.head.appendChild(script);
    return () => script.remove();
  }, [json, schemaId]);
  return null;
};

<TechArticleSchema title={"Structured Output"} description={"Structured output lets you constrain LLM responses to valid JSON that follows a specific format."} path="v3/llms/structured-output" articleSection="LLMs" about={"LLM API"} proficiencyLevel="Intermediate" keywords={["Eden AI", "AI API", "LLM API", "chat completion", "OpenAI compatible"]} datePublished="2026-05-06T00:00:00Z" dateModified="2026-05-07T00:00:00Z" />

Structured output lets you constrain LLM responses to valid JSON that follows a specific format. This is useful when you need to parse model output programmatically -- for example, extracting entities, generating API payloads, or building data pipelines.

## Overview

Eden AI supports two modes of structured output through the `response_format` parameter:

| Mode          | Description                                                                   |
| ------------- | ----------------------------------------------------------------------------- |
| `json_object` | The model returns valid JSON. You guide the structure via your system prompt. |
| `json_schema` | The model returns JSON that conforms to a specific JSON Schema you provide.   |

<Note>
  When using `response_format`, always include a system or user message that instructs the model to respond in JSON. Some providers require this instruction to be present.
</Note>

## JSON Object Mode

The simplest approach: set `response_format` to `{"type": "json_object"}` and instruct the model to return JSON in your prompt.

<CodeGroup>
  ```python Python theme={null}
  import requests
  import json

  url = "https://api.edenai.run/v3/chat/completions"
  headers = {
      "Authorization": "Bearer YOUR_API_KEY",
      "Content-Type": "application/json"
  }

  payload = {
      "model": "openai/gpt-4o",
      "response_format": {"type": "json_object"},
      "messages": [
          {
              "role": "system",
              "content": "You extract contact information. Respond with a JSON object containing: name, email, phone, company. Use null for missing fields."
          },
          {
              "role": "user",
              "content": "John Smith works at Acme Corp. His email is john@acme.com and his phone is 555-0123."
          }
      ]
  }

  response = requests.post(url, headers=headers, json=payload)
  result = response.json()

  # Parse the JSON response
  content = result["choices"][0]["message"]["content"]
  data = json.loads(content)
  print(json.dumps(data, indent=2))
  ```

  ```bash cURL theme={null}
  curl -X POST https://api.edenai.run/v3/chat/completions \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "openai/gpt-4o",
      "response_format": {"type": "json_object"},
      "messages": [
        {
          "role": "system",
          "content": "You extract contact information. Respond with a JSON object containing: name, email, phone, company. Use null for missing fields."
        },
        {
          "role": "user",
          "content": "John Smith works at Acme Corp. His email is john@acme.com and his phone is 555-0123."
        }
      ]
    }'
  ```
</CodeGroup>

**Example response:**

```json theme={null}
{
  "name": "John Smith",
  "email": "john@acme.com",
  "phone": "555-0123",
  "company": "Acme Corp"
}
```

## JSON Schema Mode

For stricter control, provide a JSON Schema that the model's response must conform to. This guarantees the output matches your expected structure.

<CodeGroup>
  ```python Python theme={null}
  import requests
  import json

  url = "https://api.edenai.run/v3/chat/completions"
  headers = {
      "Authorization": "Bearer YOUR_API_KEY",
      "Content-Type": "application/json"
  }

  payload = {
      "model": "openai/gpt-4o",
      "response_format": {
          "type": "json_schema",
          "json_schema": {
              "name": "product_review",
              "strict": True,
              "schema": {
                  "type": "object",
                  "properties": {
                      "sentiment": {
                          "type": "string",
                          "enum": ["positive", "negative", "neutral"]
                      },
                      "rating": {
                          "type": "integer",
                          "description": "Rating from 1 to 5"
                      },
                      "summary": {
                          "type": "string",
                          "description": "One-sentence summary of the review"
                      },
                      "keywords": {
                          "type": "array",
                          "items": {"type": "string"},
                          "description": "Key topics mentioned"
                      }
                  },
                  "required": ["sentiment", "rating", "summary", "keywords"],
                  "additionalProperties": False
              }
          }
      },
      "messages": [
          {
              "role": "system",
              "content": "Analyze the following product review and return structured data."
          },
          {
              "role": "user",
              "content": "This laptop is amazing! Great battery life, super fast processor, and the screen is gorgeous. Only downside is it's a bit heavy. Definitely worth the price."
          }
      ]
  }

  response = requests.post(url, headers=headers, json=payload)
  result = response.json()

  content = result["choices"][0]["message"]["content"]
  data = json.loads(content)
  print(json.dumps(data, indent=2))
  ```

  ```bash cURL theme={null}
  curl -X POST https://api.edenai.run/v3/chat/completions \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "openai/gpt-4o",
      "response_format": {
        "type": "json_schema",
        "json_schema": {
          "name": "product_review",
          "strict": true,
          "schema": {
            "type": "object",
            "properties": {
              "sentiment": {"type": "string", "enum": ["positive", "negative", "neutral"]},
              "rating": {"type": "integer"},
              "summary": {"type": "string"},
              "keywords": {"type": "array", "items": {"type": "string"}}
            },
            "required": ["sentiment", "rating", "summary", "keywords"],
            "additionalProperties": false
          }
        }
      },
      "messages": [
        {"role": "system", "content": "Analyze the following product review and return structured data."},
        {"role": "user", "content": "This laptop is amazing! Great battery life, super fast processor, and the screen is gorgeous. Only downside is it'\''s a bit heavy. Definitely worth the price."}
      ]
    }'
  ```
</CodeGroup>

**Example response:**

```json theme={null}
{
  "sentiment": "positive",
  "rating": 4,
  "summary": "An excellent laptop with great performance and display, though slightly heavy.",
  "keywords": ["battery life", "processor", "screen", "heavy", "price"]
}
```

## Supported Providers

| Provider                          | `json_object` | `json_schema` |
| --------------------------------- | :-----------: | :-----------: |
| OpenAI (GPT-4o, GPT-4 Turbo)      |      Yes      |      Yes      |
| Anthropic (Claude 3.5+, Claude 4) |      Yes      |      Yes      |
| Google (Gemini 1.5+, Gemini 2.5)  |      Yes      |      Yes      |

<Info>
  Provider support for structured output varies. If a provider does not support `json_schema`, use `json_object` mode with a detailed system prompt describing the expected format.
</Info>

## Best Practices

* **Always include JSON instructions in your prompt.** Even with `response_format` set, some providers require the prompt to mention JSON output.
* **Use `json_schema` for critical parsing.** When your downstream code depends on exact field names and types, `json_schema` mode is more reliable than `json_object`.
* **Set `additionalProperties: false`** in your schema to prevent the model from adding extra fields.
* **Always parse the response.** The `content` field is still a string -- use `json.loads()` (Python) or `JSON.parse()` (JavaScript) to convert it to a structured object.
* **Handle parsing errors gracefully.** In rare cases, the model may return malformed JSON. Wrap your parsing in a try/catch block.

## Next Steps

<CardGroup cols={2}>
  <Card title="LLMs vs Expert Models" icon="book-open" href="/v3/overview/llms-vs-expert-models">
    When to use structured LLM output vs Expert Model schemas
  </Card>
</CardGroup>