The short version
Open benchmark for deterministic structured outputs across text, image, and audio.
From the HackerLinks archive
Structured Output Benchmark
Open benchmark for deterministic structured outputs across text, image, and audio.
At a glance:
First seen:2026-04-29
Last seen:2026-04-29
Times seen:1
Website:interfaze.ai
Why it caught our attention
It targets a real agent failure mode: valid JSON with wrong values.
Where it surfaced on Hacker News
2026-04-29
Editorial paraphrase
Commenters focused on value accuracy, modality-specific rankings, and the need to measure structured hallucinations.
Original threadShow HN: A new benchmark for testing LLMs for deterministic outputs