Acropolis

o3

OpenAIGrade A. Evidence grade A of A to D: strongest.ProprietaryReleased 16 Apr 2025Consistent across benchmarks
Compare with peers

Domain profile

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

  • Agentic (3 boards): 17 (scale 0 to 100)Agentic (3 boards)17
  • Coding (2 boards): 35 (scale 0 to 100)Coding (2 boards)35
  • Finance (1 boards): 40 (scale 0 to 100)Finance (1 boards)40
  • General text (5 boards): 58 (scale 0 to 100)General text (5 boards)58
  • Math (2 boards): 57 (scale 0 to 100)Math (2 boards)57
  • Science (3 boards): 58 (scale 0 to 100)Science (3 boards)58
  • Search (2 boards): 36 (scale 0 to 100)Search (2 boards)36
  • Vision (3 boards): 51 (scale 0 to 100)Vision (3 boards)51

Benchmark appearances

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

BenchmarkRankScoreCostStatusObserved
GDPval

Epoch AI Benchmarking Hub · medium

#7 / 11
30.8%UnknownStandardNot stated
METR time horizon

METR time horizons · triframe_inspect/triframe_agent

#11 / 23Tied #1
119.7 min[74.6 min to 190.9 min]UnknownStandardNot stated
LMArena Vision (style control)

LMArena · Entity recognition · Default

#16 / 52
1,231[1,199 to 1,262]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Captioning · Default

#19 / 34
1,211[1,183 to 1,239]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Creative writing · Default

#19 / 34
1,197[1,183 to 1,211]UnknownStandard9 Jan 2026
LMArena Vision (style control)

LMArena · Captioning · Default

#19 / 34
1,184[1,154 to 1,214]UnknownStandard9 Nov 2026
LMArena Search (style control)

LMArena · Overall · search

#20 / 42
1,197[1,190 to 1,204]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Creative writing · Default

#20 / 34
1,192[1,177 to 1,207]UnknownStandard9 Jan 2026
HLE

Epoch AI Benchmarking Hub · high

#21 / 56
20.3%[17.2% to 23.4%]UnknownStandardNot stated
HLE

Epoch AI Benchmarking Hub · medium

#23 / 56
19.2%[16.2% to 22.2%]UnknownStandardNot stated
LMArena Vision

LMArena · Entity recognition · Default

#24 / 52
1,233[1,203 to 1,263]UnknownStandard9 Nov 2026
OSWorld

Epoch AI Benchmarking Hub · medium

#30 / 38
23%UnknownStandardNot stated
SWE-Bench verified

Epoch AI Benchmarking Hub · medium

#30 / 35
62.3%[58% to 66.6%]UnknownStandard12 Feb 2026
OSWorld

Epoch AI Benchmarking Hub · medium

#31 / 38
17.2%UnknownStandardNot stated
LMArena Search

LMArena · Overall · search

#34 / 42
1,145[1,139 to 1,150]UnknownStandard9 Nov 2026
OSWorld

Epoch AI Benchmarking Hub · medium

#34 / 38
9.1%UnknownStandardNot stated
FrontierMath-2025-02-28-Private

Epoch AI Benchmarking Hub · high

#45 / 101
18.7%[14.2% to 23.2%]UnknownStandard16 Nov 2025
FrontierMath-2025-02-28-Private

Epoch AI Benchmarking Hub · medium

#47 / 101
16.9%[12.6% to 21.2%]UnknownStandard16 Nov 2025
FrontierMath-2025-02-28-Private

Epoch AI Benchmarking Hub · low

#56 / 101
9.7%[6.3% to 13.1%]UnknownStandard17 Nov 2025
LMArena Text (style control)

LMArena · Polish · Default

#67 / 243
1,458[1,447 to 1,469]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry medicine and healthcare · Default

#69 / 391
1,476[1,465 to 1,487]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Japanese · Default

#69 / 288
1,424[1,407 to 1,442]UnknownStandard11 Nov 2026
LMArena Vision (style control)

LMArena · Humor · Default

#69 / 102
1,210[1,190 to 1,231]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Humor · Default

#71 / 102
1,221[1,202 to 1,241]UnknownStandard9 Nov 2026
LMArena Text

LMArena · Polish · Default

#74 / 243
1,438[1,427 to 1,449]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Japanese · Default

#77 / 288
1,404[1,386 to 1,421]UnknownStandard11 Nov 2026
LMArena Vision (style control)

LMArena · Homework · Default

#79 / 118
1,250[1,235 to 1,265]UnknownStandard9 Nov 2026
LMArena Text

LMArena · Industry medicine and healthcare · Default

#81 / 391
1,448[1,437 to 1,459]UnknownStandard11 Nov 2026
LMArena Vision

LMArena · Creative writing vision · Default

#81 / 106
1,201[1,183 to 1,220]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Chinese · Default

#82 / 135
1,234[1,217 to 1,250]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Overall · Default

#82 / 167
1,216[1,209 to 1,222]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · English · Default

#83 / 167
1,218[1,209 to 1,227]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Chinese · Default

#84 / 135
1,237[1,221 to 1,253]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Creative writing vision · Default

#84 / 106
1,194[1,175 to 1,212]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · OCR · Default

#84 / 123
1,226[1,217 to 1,234]UnknownStandard9 Nov 2026
LMArena Text (style control)

LMArena · Math · Default

#85 / 404
1,448[1,438 to 1,458]UnknownStandard11 Nov 2026
LMArena Vision

LMArena · Homework · Default

#85 / 118
1,235[1,220 to 1,250]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Overall · Default

#86 / 167
1,214[1,207 to 1,220]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Diagram · Default

#87 / 123
1,233[1,221 to 1,245]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · English · Default

#88 / 167
1,221[1,212 to 1,229]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Diagram · Default

#89 / 123
1,219[1,207 to 1,231]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · OCR · Default

#89 / 123
1,218[1,210 to 1,226]UnknownStandard9 Nov 2026
LMArena Text (style control)

LMArena · Korean · Default

#91 / 289
1,394[1,375 to 1,412]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · German · Default

#95 / 320
1,439[1,423 to 1,455]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry legal and government · Default

#95 / 393
1,452[1,442 to 1,463]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · French · Default

#97 / 301
1,460[1,438 to 1,482]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry mathematical · Default

#98 / 402
1,449[1,438 to 1,459]UnknownStandard11 Nov 2026
LMArena Text

LMArena · German · Default

#100 / 320
1,420[1,404 to 1,436]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry life and physical and social science · Default

#103 / 419
1,456[1,449 to 1,463]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Non English · Default

#108 / 421
1,422[1,418 to 1,427]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Russian · Default

#111 / 385
1,429[1,421 to 1,438]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Math · Default

#112 / 404
1,426[1,416 to 1,436]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Korean · Default

#113 / 289
1,370[1,352 to 1,388]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Overall · Default

#114 / 421
1,432[1,428 to 1,435]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry legal and government · Default

#115 / 393
1,429[1,418 to 1,439]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Exclude ties · Default

#115 / 421
1,425[1,420 to 1,430]UnknownStandard11 Nov 2026
LMArena Text

LMArena · French · Default

#119 / 301
1,430[1,408 to 1,452]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · English · Default

#121 / 421
1,439[1,435 to 1,444]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry business and management and financial operations · Default

#121 / 414
1,427[1,420 to 1,434]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Spanish · Default

#123 / 304
1,417[1,399 to 1,435]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry entertainment and sports and media · Default

#124 / 419
1,394[1,388 to 1,401]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Chinese · Default

#125 / 395
1,462[1,452 to 1,472]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Expert · Default

#125 / 371
1,448[1,437 to 1,460]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Russian · Default

#127 / 385
1,406[1,398 to 1,415]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry life and physical and social science · Default

#128 / 419
1,430[1,423 to 1,436]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Hard prompts English · Default

#129 / 419
1,450[1,444 to 1,456]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry mathematical · Default

#132 / 402
1,423[1,413 to 1,433]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Non English · Default

#132 / 421
1,401[1,397 to 1,406]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Hard prompts · Default

#134 / 421
1,441[1,436 to 1,446]UnknownStandard11 Nov 2026
GPQA diamond

Epoch AI Benchmarking Hub · high

#137 / 338
81.8%[77.6% to 86%]UnknownStandard16 Apr 2025
LMArena Text (style control)

LMArena · Industry software and IT services · Default

#137 / 421
1,455[1,450 to 1,460]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry writing and literature and language · Default

#137 / 420
1,396[1,390 to 1,402]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Multi turn · Default

#137 / 419
1,420[1,414 to 1,427]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry entertainment and sports and media · Default

#138 / 419
1,373[1,366 to 1,379]UnknownStandard11 Nov 2026
GPQA diamond

Epoch AI Benchmarking Hub · medium

#143 / 338
80.8%[75.3% to 86.3%]UnknownStandard7 Aug 2026
LMArena Text

LMArena · Exclude ties · Default

#143 / 421
1,393[1,388 to 1,398]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Overall · Default

#143 / 421
1,410[1,406 to 1,413]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Coding · Default

#143 / 416
1,460[1,454 to 1,466]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Creative writing · Default

#143 / 419
1,383[1,376 to 1,390]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Chinese · Default

#146 / 395
1,437[1,427 to 1,447]UnknownStandard11 Nov 2026
GPQA diamond

Epoch AI Benchmarking Hub · low

#148 / 338
79.8%[74.2% to 85.4%]UnknownStandard15 Jul 2026
LMArena Text

LMArena · English · Default

#148 / 421
1,416[1,412 to 1,421]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Instruction following · Default

#148 / 421
1,403[1,397 to 1,408]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Multi turn · Default

#149 / 419
1,405[1,398 to 1,412]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry business and management and financial operations · Default

#156 / 414
1,392[1,386 to 1,399]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Spanish · Default

#156 / 304
1,395[1,377 to 1,413]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Expert · Default

#158 / 371
1,402[1,390 to 1,413]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Creative writing · Default

#159 / 419
1,359[1,351 to 1,366]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Longer query · Default

#160 / 399
1,411[1,405 to 1,417]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry writing and literature and language · Default

#161 / 420
1,371[1,365 to 1,377]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Hard prompts English · Default

#163 / 419
1,410[1,404 to 1,417]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry software and IT services · Default

#165 / 421
1,415[1,409 to 1,420]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Hard prompts · Default

#166 / 421
1,402[1,397 to 1,407]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Coding · Default

#171 / 416
1,407[1,401 to 1,414]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Instruction following · Default

#174 / 421
1,368[1,362 to 1,373]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Longer query · Default

#185 / 399
1,372[1,365 to 1,378]UnknownStandard11 Nov 2026
ARC-AGI-2

Epoch AI Benchmarking Hub · high

#193 / 299
6.5%$0.83StandardNot stated
ARC-AGI-2

Epoch AI Benchmarking Hub · medium

#223 / 299
3%$0.48StandardNot stated
ARC-AGI-2

Epoch AI Benchmarking Hub · low

#239 / 299
2%$0.23StandardNot stated

Reasoning effort scaling

Same benchmark and harness measured across multiple reasoning effort settings.

  • ARC-AGI-2

    Epoch AI Benchmarking Hub

    Monotone. Evidence grade A of A to D: strongest.
    low

    2.0

    $0.23

    medium

    3.0

    $0.48

    high

    6.5

    $0.83

  • FrontierMath-2025-02-28-Private

    Epoch AI Benchmarking Hub

    Monotone. Evidence grade A of A to D: strongest.
    low

    9.7

    medium

    16.9

    high

    18.7

  • GPQA diamond

    Epoch AI Benchmarking Hub

    Monotone. Evidence grade A of A to D: strongest.
    low

    79.8

    medium

    80.8

    high

    81.8

  • HLE

    Epoch AI Benchmarking Hub

    Monotone. Evidence grade A of A to D: strongest.
    medium

    19.2

    high

    20.3

Closest head-to-head rivals

Direct pairwise win rates for o3 across shared benchmark boards.