Acropolis

Gemini 2.5 Pro Preview (06-05)

GoogleGrade C. Evidence grade C of A to D: thin.ProprietaryReleased 5 Jun 2025Consistent across benchmarks
Compare with peers

Domain profile

Median within-board percentile by capability domain across boards with sufficient peer overlap.

  • General text (2 boards): 32 (scale 0 to 100)General text (2 boards)32
  • Math (2 boards): 54 (scale 0 to 100)Math (2 boards)54
  • Science (3 boards): 60 (scale 0 to 100)Science (3 boards)60

Benchmark appearances

Best verified run per board, with confidence intervals and Pareto frontier status.

BenchmarkRankScoreCostStatusObserved
HLE

Epoch AI Benchmarking Hub · Default

#20 / 56
21.6%[18.5% to 24.8%]UnknownStandardNot stated
FrontierMath-2025-02-28-Private

Epoch AI Benchmarking Hub · Default

#54 / 101
10.3%[6.8% to 13.9%]UnknownStandard3 Jul 2025
GPQA diamond

Epoch AI Benchmarking Hub · Default

#111 / 338
84.8%[79.8% to 89.9%]UnknownStandard5 Jun 2025
ARC-AGI-2

Epoch AI Benchmarking Hub · Default

#281 / 299
0%$0.09StandardNot stated

Closest head-to-head rivals

Direct pairwise win rates for Gemini 2.5 Pro Preview (06-05) across shared benchmark boards.