Benchmarks
| Benchmark | Claude Sonnet 5.5 | GPT-6 Sol | Difference |
|---|
| FrontierCode 1.1 (Main) | 52.1%Xhigh effort. 46.2% at Max | 49.3% | Sonnet +2.8 pointsSol +3.1 points vs Sonnet at Max |
| GDPval-AA v2.1 | 1844 Elo | 1487 Elo | Sonnet +357 Elo |
| AA-Briefcase v1.1 | 1811 Elo | 1483 Elo | Sonnet +328 Elo |
| Chartography, no tools | 61.6% | 53.6% | Sonnet +8.0 points |
Artificial Analysis benchmarks
| Benchmark | Claude Sonnet 5.5 (Max) | GPT-6 Sol (Max) | Difference |
|---|
| Intelligence Index v4.3.2 | 56 | 48 | Sonnet +8 |
| AA-Briefcase v1.1 | 1811 Elo | 1483 Elo | Sonnet +328 Elo |
| GDPval-AA v2.1 | 1844 Elo | 1487 Elo | Sonnet +357 Elo |
| AutomationBench-AA | 71% | 62% | Sonnet +9 points |
| Terminal-Bench 4.0 | 64% | 44% | Sonnet +20 points |
| SciCode | 61% | 58% | Sonnet +3 points |
| Humanity's Last Exam | 55% | 48% | Sonnet +7 points |
| CritPt, physics | 31% | 31% | Tie |
| AA-Omniscience | 32 | 27 | Sonnet +5 |
| AA-LCR v1.1, long context | 83% | 84% | Sol +1 point |
