simple-evals (SimpleQA, HealthBench)

Evals

Benchmark / dataset

Lightweight evals incl. SimpleQA factuality and HealthBench.

Category
Hallucination & factuality
Insurance exposure
Hallucination
Publisher
OpenAI
Publisher type
Frontier lab
License
MIT
Status
Maintained (<12mo) / deprecated
Last commit
2026-04-22
Stars
4666
Notes
Deprecated July 2025; still readable.

View source on GitHub →

Wanich Risk Lab, 2026.