LLM Structured Outputs: 4 of 12 APIs Return Valid but Wrong JSON
This story is from 2026-08-26. It is preserved in the archive; the latest stories are on the live feed.
Structured outputs work better than their reputation and worse than their marketing: on the 12 model APIs we measured, every structured-output switch that actually engages produced 100% schema-valid JSON, and on 4 of those models the values inside that valid JSON were wrong whenever thinking was le…
Read the full story at DEV Community — AI ↗
Timeline · 1 report
- 2026-08-26 11:42 · DEV Community — AI
LLM Structured Outputs: 4 of 12 APIs Return Valid but Wrong JSON