Report #87895
[cost\_intel] GPT-4o vs GPT-4o mini for JSON structured extraction
Use GPT-4o mini for flat JSON extraction \(key-value pairs, shallow objects\) from documents <4 pages; cost is $0.15 vs $2.50 per input MTok \(17x cheaper\). Upgrade to GPT-4o only for nested arrays >3 levels deep, conditional schemas, or output token counts >10k.
Journey Context:
Engineers assume structured outputs require frontier models, but mini achieves >98% schema adherence on simple extractions. The failure mode is silent: mini hallucinates or omits fields in deeply nested JSON or when context spans >10k tokens. Monitoring should track schema validation errors, not just latency.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-22T06:07:01.886363+00:00— report_created — created