JSON Mode (Structured Output) — AI Glossary

Reference details

Related terms

• Function Calling
• Guardrails
• Output Parser

Example: Valid JSON is not the same as the right JSON

Asked to extract an invoice total, the model returns {"total": 0} — perfectly valid, schema-conformant, and wrong, because the figure was in a footer it did not read. JSON mode removed the parsing failure and left the extraction failure untouched. Those are different problems and only one of them is solved here.

Commonly confused with

JSON mode and structured output are often used interchangeably but differ in strength. Plain JSON mode guarantees only that the output parses; schema-constrained output additionally guarantees the fields and types you declared. If you need a specific shape, asking for JSON is not enough.

When to use it

Reach for it when:

• The output is consumed by code rather than read by a person
• You need a parse failure to be impossible rather than rare
• Extraction into typed records

Reach for something else when:

• The reader is human — schemas make prose worse
• You are treating conformance as correctness and skipping evaluation
• The schema is deeply nested, where quality falls and latency rises

Referenced by

This term is used by the following ModelRefs references:

• function calling basic
• json schema tool call

Continue your research

Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to JSON Mode (Structured Output) — AI Glossary.

• Browse models
• Review providers
• Inspect benchmarks
• Explore workflows
• Learn AI concepts
• Read decision guides
• Review trust and methodology
• Start a decision path