Acropolis

GPT-5 mini

OpenAIGrade B. Evidence grade B of A to D: good.ProprietaryReleased 7 Aug 2025Consistent across benchmarks
Compare with peers

Domain profile

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

  • Agentic (3 boards): 19 (scale 0 to 100)Agentic (3 boards)19
  • Coding (2 boards): 27 (scale 0 to 100)Coding (2 boards)27
  • Finance (1 boards): 0 (scale 0 to 100)Finance (1 boards)0
  • General text (4 boards): 57 (scale 0 to 100)General text (4 boards)57
  • Math (2 boards): 63 (scale 0 to 100)Math (2 boards)63
  • Science (3 boards): 51 (scale 0 to 100)Science (3 boards)51
  • Vision (2 boards): 44 (scale 0 to 100)Vision (2 boards)44

Benchmark appearances

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

BenchmarkRankScoreCostStatusObserved
LMArena Vision

LMArena · Captioning · high

#17 / 34
1,216[1,179 to 1,253]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Captioning · high

#21 / 34
1,173[1,130 to 1,216]UnknownStandard9 Nov 2026
HLE

Epoch AI Benchmarking Hub · Default

#22 / 56
19.4%[16.4% to 22.5%]UnknownStandardNot stated
LMArena Vision

LMArena · Creative writing · Default

#26 / 34
1,181[1,163 to 1,199]UnknownStandard9 Jan 2026
FrontierMath-2025-02-28-Private

Epoch AI Benchmarking Hub · high

#28 / 101
27.2%[22.1% to 32.4%]UnknownStandard13 Nov 2025
LMArena Vision (style control)

LMArena · Entity recognition · high

#28 / 52
1,212[1,172 to 1,253]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Creative writing · Default

#29 / 34
1,152[1,132 to 1,173]UnknownStandard9 Jan 2026
SWE-Bench verified

Epoch AI Benchmarking Hub · medium

#29 / 35
64.7%[60.4% to 68.9%]UnknownStandard1 Feb 2026
LMArena Vision

LMArena · Entity recognition · high

#31 / 52
1,226[1,190 to 1,262]UnknownStandard9 Nov 2026
FrontierMath-2025-02-28-Private

Epoch AI Benchmarking Hub · medium

#41 / 101
20.3%[15.7% to 25%]UnknownStandard13 Nov 2025
Vending-Bench 2

Epoch AI Benchmarking Hub · Default

#72 / 72
$-31.18UnknownStandardNot stated
LMArena Vision

LMArena · Humor · high

#79 / 102
1,205[1,182 to 1,227]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Humor · high

#85 / 102
1,182[1,158 to 1,206]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Homework · high

#88 / 118
1,238[1,219 to 1,258]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · English · high

#91 / 167
1,216[1,205 to 1,226]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Overall · high

#93 / 167
1,202[1,195 to 1,210]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Creative writing vision · high

#94 / 106
1,182[1,160 to 1,204]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · OCR · high

#94 / 123
1,199[1,188 to 1,209]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Chinese · high

#95 / 135
1,214[1,195 to 1,232]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Diagram · high

#95 / 123
1,202[1,187 to 1,217]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Diagram · high

#95 / 123
1,213[1,198 to 1,229]UnknownStandard9 Nov 2026
LMArena Vision

LMArena · Homework · high

#97 / 118
1,212[1,192 to 1,231]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · OCR · high

#97 / 123
1,200[1,189 to 1,211]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · English · high

#98 / 167
1,194[1,183 to 1,205]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Chinese · high

#100 / 135
1,181[1,161 to 1,201]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Creative writing vision · high

#100 / 106
1,152[1,130 to 1,175]UnknownStandard9 Nov 2026
LMArena Vision (style control)

LMArena · Overall · high

#103 / 167
1,185[1,177 to 1,193]UnknownStandard9 Nov 2026
Terminal Bench

Epoch AI Benchmarking Hub · spoox-o-m

#121 / 202
34.8%[29.5% to 40.2%]UnknownStandard7 Aug 2025
Terminal Bench

Epoch AI Benchmarking Hub · spoox-m

#123 / 202
34.8%[29.5% to 40.1%]UnknownStandard24 Dec 2025
LMArena Text (style control)

LMArena · Japanese · high

#130 / 288
1,365[1,336 to 1,393]UnknownStandard11 Nov 2026
Terminal Bench

Epoch AI Benchmarking Hub · medium · Codex CLI

#131 / 202
31.9%[26% to 37.8%]UnknownStandard4 Nov 2025
Terminal Bench

Epoch AI Benchmarking Hub · Codex CLI

#132 / 202
31.9%[26% to 37.7%]UnknownStandard7 Aug 2025
Terminal Bench

Epoch AI Benchmarking Hub · OpenHands

#136 / 202
29.2%[23.8% to 34.6%]UnknownStandard7 Aug 2025
LMArena Text (style control)

LMArena · Polish · high

#137 / 243
1,400[1,384 to 1,416]UnknownStandard11 Nov 2026
Terminal Bench

Epoch AI Benchmarking Hub · medium · OpenHands

#138 / 202
29.2%[23.7% to 34.7%]UnknownStandard2 Nov 2025
LMArena Text

LMArena · Japanese · high

#140 / 288
1,341[1,313 to 1,370]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Polish · high

#150 / 243
1,378[1,363 to 1,394]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · German · high

#152 / 320
1,385[1,360 to 1,410]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Math · high

#154 / 404
1,407[1,392 to 1,423]UnknownStandard11 Nov 2026
Terminal Bench

Epoch AI Benchmarking Hub · Terminus 2

#161 / 202
24%[19.1% to 29%]UnknownStandard7 Aug 2025
LMArena Text (style control)

LMArena · French · high

#162 / 301
1,410[1,379 to 1,440]UnknownStandard11 Nov 2026
Terminal Bench

Epoch AI Benchmarking Hub · medium · Terminus 2

#162 / 202
24%[19.1% to 28.9%]UnknownStandard31 Oct 2025
LMArena Text

LMArena · German · high

#164 / 320
1,365[1,341 to 1,389]UnknownStandard11 Nov 2026
LMArena Text

LMArena · French · high

#166 / 301
1,386[1,356 to 1,416]UnknownStandard11 Nov 2026
Terminal Bench

Epoch AI Benchmarking Hub · Mini-SWE-Agent

#167 / 202
22.2%[17.1% to 27.4%]UnknownStandard7 Aug 2025
Terminal Bench

Epoch AI Benchmarking Hub · medium · Mini-SWE-Agent

#168 / 202
22.2%[17.1% to 27.3%]UnknownStandard3 Nov 2025
LMArena Text (style control)

LMArena · Industry mathematical · high

#169 / 402
1,409[1,392 to 1,425]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Coding · high

#173 / 416
1,406[1,398 to 1,415]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Korean · high

#175 / 289
1,304[1,280 to 1,328]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Non English · high

#175 / 421
1,378[1,372 to 1,384]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Korean · high

#177 / 289
1,321[1,296 to 1,346]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Russian · high

#178 / 385
1,378[1,365 to 1,392]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Expert · high

#179 / 371
1,379[1,362 to 1,396]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Expert · high

#179 / 371
1,403[1,386 to 1,421]UnknownStandard11 Nov 2026
GPQA diamond

Epoch AI Benchmarking Hub · high

#180 / 338
75%[70.5% to 79.5%]UnknownStandard30 Oct 2025
LMArena Text

LMArena · Hard prompts · high

#180 / 421
1,380[1,374 to 1,386]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Non English · high

#180 / 421
1,363[1,357 to 1,368]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry medicine and healthcare · high

#180 / 391
1,408[1,392 to 1,424]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry legal and government · high

#181 / 393
1,397[1,382 to 1,411]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry software and IT services · high

#182 / 421
1,394[1,387 to 1,401]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Overall · high

#182 / 421
1,390[1,385 to 1,395]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry mathematical · high

#183 / 402
1,381[1,365 to 1,398]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Exclude ties · high

#183 / 421
1,366[1,360 to 1,373]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry entertainment and sports and media · high

#184 / 419
1,340[1,332 to 1,349]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Instruction following · high

#184 / 421
1,357[1,349 to 1,364]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Math · high

#185 / 404
1,378[1,363 to 1,394]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Overall · high

#185 / 421
1,373[1,369 to 1,378]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Russian · high

#185 / 385
1,362[1,349 to 1,375]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry legal and government · high

#186 / 393
1,375[1,360 to 1,389]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry medicine and healthcare · high

#186 / 391
1,380[1,364 to 1,395]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry entertainment and sports and media · high

#186 / 419
1,348[1,339 to 1,357]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Spanish · high

#186 / 304
1,373[1,351 to 1,394]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Exclude ties · high

#187 / 421
1,343[1,336 to 1,349]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Chinese · high

#187 / 395
1,404[1,390 to 1,418]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry business and management and financial operations · high

#187 / 414
1,382[1,373 to 1,391]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Hard prompts English · high

#188 / 419
1,383[1,375 to 1,392]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Coding · high

#188 / 416
1,432[1,424 to 1,441]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Hard prompts · high

#188 / 421
1,404[1,398 to 1,410]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry business and management and financial operations · high

#189 / 414
1,358[1,349 to 1,367]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · English · high

#189 / 421
1,398[1,392 to 1,404]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Spanish · high

#190 / 304
1,355[1,334 to 1,376]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Instruction following · high

#191 / 421
1,372[1,365 to 1,380]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Hard prompts English · high

#192 / 419
1,410[1,401 to 1,418]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry life and physical and social science · high

#192 / 419
1,398[1,388 to 1,407]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry software and IT services · high

#192 / 421
1,416[1,409 to 1,423]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Chinese · high

#193 / 395
1,386[1,372 to 1,399]UnknownStandard11 Nov 2026
GPQA diamond

Epoch AI Benchmarking Hub · minimal

#194 / 338
71.7%[65.4% to 78%]UnknownStandard7 Aug 2026
LMArena Text

LMArena · English · high

#195 / 421
1,379[1,373 to 1,385]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry writing and literature and language · high

#195 / 420
1,344[1,336 to 1,352]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Multi turn · high

#195 / 419
1,363[1,353 to 1,372]UnknownStandard11 Nov 2026
GPQA diamond

Epoch AI Benchmarking Hub · medium

#197 / 338
71.7%[67.1% to 76.2%]UnknownStandard7 Aug 2025
LMArena Text

LMArena · Longer query · high

#197 / 399
1,355[1,347 to 1,363]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Multi turn · high

#198 / 419
1,373[1,363 to 1,382]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Creative writing · high

#200 / 419
1,325[1,315 to 1,335]UnknownStandard11 Nov 2026
LMArena Text

LMArena · Industry life and physical and social science · high

#200 / 419
1,370[1,361 to 1,379]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Industry writing and literature and language · high

#200 / 420
1,351[1,343 to 1,359]UnknownStandard11 Nov 2026
LMArena Text (style control)

LMArena · Longer query · high

#205 / 399
1,370[1,362 to 1,379]UnknownStandard11 Nov 2026
ARC-AGI-2

Epoch AI Benchmarking Hub · high

#212 / 299
4.4%$0.20StandardNot stated
ARC-AGI-2

Epoch AI Benchmarking Hub · medium

#214 / 299
4%$0.06StandardNot stated
LMArena Text (style control)

LMArena · Creative writing · high

#216 / 419
1,323[1,313 to 1,333]UnknownStandard11 Nov 2026
ARC-AGI-2

Epoch AI Benchmarking Hub · minimal

#247 / 299
1.7%$0.01On frontier. Evidence grade A of A to D: strongest.Not stated
ARC-AGI-2

Epoch AI Benchmarking Hub · low

#271 / 299
0.8%$0.02StandardNot stated

Reasoning effort scaling

Same benchmark and harness measured across multiple reasoning effort settings.

  • ARC-AGI-2

    Epoch AI Benchmarking Hub

    1 regression. Evidence grade D of A to D: weakest.
    minimal

    1.7

    $0.01

    low

    0.8

    $0.02

    medium

    4.0

    $0.06

    high

    4.4

    $0.20

  • FrontierMath-2025-02-28-Private

    Epoch AI Benchmarking Hub

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

    20.3

    high

    27.2

  • GPQA diamond

    Epoch AI Benchmarking Hub

    1 regression. Evidence grade D of A to D: weakest.
    minimal

    71.7

    medium

    71.7

    high

    75.0

Closest head-to-head rivals

Direct pairwise win rates for GPT-5 mini across shared benchmark boards.