# Benchmark explorer

> The best AI score over time on five hard tests, what each measures, and how it compares with people.

Benchmark data from Epoch AI (CC BY) and official leaderboards.

| Benchmark | Best score today | Change | Details |
| --- | --- | --- | --- |
| GPQA Diamond | 96% | +3 pts in 12 months | https://www.superintelligence.com.au/benchmarks/gpqa-diamond.md |
| SWE-bench Verified | 84% | +7 pts in 12 months | https://www.superintelligence.com.au/benchmarks/swe-bench-verified.md |
| FrontierMath | 94% | +20 pts in 12 months | https://www.superintelligence.com.au/benchmarks/frontiermath.md |
| Humanity’s Last Exam | 55% | +17 pts in 12 months | https://www.superintelligence.com.au/benchmarks/humanitys-last-exam.md |
| ARC-AGI-2 | 95% | +41 pts in 12 months | https://www.superintelligence.com.au/benchmarks/arc-agi-2.md |

Download: https://www.superintelligence.com.au/data/benchmarks.csv