Atlas

Benchmarks

← All benchmarks

Global PIQA - Parallel (Exact Match)

General QA · 2025-10-28

Global PIQA's parallel split presents aligned physical-commonsense questions with four answer options. This definition scores generated answers using exact match without the strict extraction filter.

Top models (higher is better)

ModelScore
Gemini 3.1 Pro Preview96.8
GPT-5.491.2
Gemini 3.1 Flash-Lite Preview88.7
Sonnet 4.687.7
GPT-5.4 Mini85.2
GPT-5.4 Nano76.3
Loading Atlas data…