Show HN: A new benchmark for testing LLMs for deterministic outputs

When building workflows that rely on LLMs, we commonly use structured output for programmatic use cases like converting an invoice into rows or meeting transcripts into tickets or even complex PDFs into database entries....