Artificial Analysis · Intelligence IndexBenchmarks, as Artificial Analysis ran them- 94%GPQA Diamond
- 42%Humanity's Last Exam
- 11%Terminal-Bench 4.0
- 82%Terminal-Bench 2.1
- 54%SciCode
- 49%τ²-Bench Banking
- 80%Long context reasoning
- Speed
- 38.8 tokens/s
- First answer
- 1.9 s
- thinking included
- Per 1M tokens
- $2.00 in · $6.00 out
Artificial Analysis, read 22 October 2026