Show HN: A new benchmark for testing LLMs for deterministic outputs 2026-04-29 · Hacker News Read full story at source