structured_output_python

Structured Output Example (Python)

Shows how to get typed responses from Subconscious. The API speaks the OpenAI Chat Completions protocol, so we point the official openai SDK at https://api.subconscious.dev/v1, pass a JSON schema via response_format, and validate the reply with a Pydantic model.

Thinking is disabled on the request (enable_thinking: false) so the model returns clean JSON with no prose preamble — a requirement for reliable structured output parsing.

Prerequisites

• Python 3.9+
• A Subconscious API key — get one at https://www.subconscious.dev/platform/api-keys

Setup

pip install .

Create a .env file:

SUBCONSCIOUS_API_KEY=your_key

Run

python main.py

Enter some text when prompted. The model will analyze the sentiment and return a structured response validated by a Pydantic model.

Expected Output

Enter a text to analyze: The new product launch went incredibly well!
Sentiment:  positive
Confidence: 0.97
Keywords:   new product launch, incredibly well