Docs Mismatch: response_format dictionary syntax conflicting with types.GenerateContentConfig

[kaveh0007]

Description

There is a syntax conflict in the official structured output documentation. The Python code blocks on the Structured outputs documentation page use a deeply nested raw dictionary structure for configuring JSON schemas.

config={
    "response_format": {"text": {"mime_type": "application/json", "schema": Recipe.model_json_schema()}},
}

However, the modern google-genai Python SDK recommends and implements structured generation using flat, top-level keys within types.GenerateContentConfig (or equivalent dictionary mapping):

Example:

config={
    response_mime_type="application/json",
    response_schema=Recipe
}

Environment details

• Programming language: Python
• OS: linux-mint
• Language runtime version: 3.12.3
• Package version: 2.6.0

Steps to reproduce

1. Navigate to the documentation at: https://ai.google.dev/gemini-api/docs/structured-output

2. Observe the Python tabs under the "Recipe Extractor", "Streaming", and "Structured outputs with tools" sections.

3. They all use the nested "response_format" dictionary configuration instead of the native SDK schema parameters.

Making sure to follow these steps will guarantee the quickest resolution possible.

Thanks!

[kaveh0007]

I would love to fix this documentation mismatch myself. Could you please assign this issue to me?

[kkorpal]

Hi @kaveh0007, Great catch! You're absolutely right, and I've escalated this to our documentation team to get those snippets updated. Thanks for helping us keep everything accurate!

[kaveh0007]

Hi @kaveh0007, Great catch! You're absolutely right, and I've escalated this to our documentation team to get those snippets updated. Thanks for helping us keep everything accurate!

Thanks for the update, @kkorpal. Glad I could help catch this!