The AI-102 exam has been retired
It was retired on June 30, 2026. It was replaced by AI-103 (Microsoft Azure AI Apps and Agents Developer Associate). Go to the AI-103 study guide →
Copyright (c) 2026 MindMesh Academy. All rights reserved. This content is proprietary and may not be reproduced or distributed without permission.
3.2.1.3. Response Control and Formatting
3.2.1.3. Response Control and Formatting
Response control parameters determine how the model structures its output. For applications that need to parse responses programmatically, structured output is essential—free-form text requires fragile regex parsing.
Key response control mechanisms:
response_format: {"type": "json_object"}: Forces valid JSON outputstopsequences: Tell the model when to stop generating (e.g.,["\n", "User:"])max_tokens: Hard limit on response length (prevents runaway costs)- Function calling: Structured tool invocation with validated parameters
Written byAlvin Varughese
Founder•18 professional certifications