Humanity Statsv1
Live
Top 73

The benchmarks with the most models scored

Ranked by how many model versions have been scored — a proxy for which evaluations the field actually treats as load-bearing.

Epoch Capabilities Index leads with 462, ahead of MMLU (217) and Terminal-Bench (203).

Benchmarks saturate: a test that separated models two years ago is often solved by everything today, which is why the set keeps growing and why an aggregate index exists at all. The Epoch Capabilities Index combines across them so progress stays visible after individual tests top out.

#NameMetricLatest modelModels scored
01Epoch Capabilities IndexECI ScoreJul 24, 2026462
02MMLUEMFeb 26, 2025217
03Terminal-BenchAccuracy meanApr 23, 2026203
04GPQA Diamondmean_scoreJul 24, 2026199
05OTIS Mock AIME 2024-2025mean_scoreJul 24, 2026173
06GSM8KEMNov 29, 2024170
07ARC-AGIScoreJul 24, 2026159
08ARC-AGI-2ScoreJul 24, 2026159
09WeirdMLAccuracyJul 9, 2026146
10WinoGrandeAccuracyDec 26, 2024137
11BoolQScoreAug 17, 2024136
12ARC AI2Challenge scoreDec 26, 2024134
13TriviaQAEMDec 26, 2024115
14PIQAScoreDec 26, 2024113
15HellaSwagOverall accuracyDec 26, 2024113
16MATH Level 5mean_scoreOct 15, 2025108
17FrontierMathmean_scoreMay 28, 2026101
18SimpleBenchScore (AVG@5)Jul 9, 202691
19BIG-Bench HardAverageDec 26, 202488
20ALE BenchPerformanceJun 16, 202687
21Chess Puzzlesmean_scoreJul 24, 202685
22WebDev ArenaArena ScoreJun 16, 202682
23SciCodeScoreJul 24, 202675
24CritptAccuracyJul 24, 202674
25Aider PolyglotPercent correctDec 1, 202572
26FrontierMath Tier 4mean_scoreMay 28, 202672
27OpenBookQAAccuracyApr 23, 202470
28SimpleQA Verifiedmean_scoreJul 24, 202668
29ForecastBenchOverall scoreMar 17, 202668
30LiveBenchGlobal averageNov 13, 202554
31Vending-Bench 2ScoreJul 9, 202654
32LAMBADAScoreNov 30, 202353
33Humanity's Last ExamAccuracyApr 16, 202651
34Apex AgentsPass@1 scoreJul 24, 202650
35METR Time Horizonsaverage_scoreApr 7, 202650
36Video-MMEOverall (no subtitles)Jun 18, 202550
37Lech Mazur WritingMean scoreAug 7, 202549
38ProofbenchAccuracyJul 24, 202644
39Enigma EvalAccuracyMar 5, 202642
40Fictionlivebench120k token scoreJan 27, 202642
41VpctCorrectDec 17, 202538
42DeepswePass@1Jul 9, 202637
43BalrogAverage progressFeb 19, 202636
44DeepresearchbenchAverage scoreMay 28, 202636
45GSOScore OPT@1Apr 23, 202636
46SWE-bench Verifiedmean_scoreJun 16, 202635
47GeobenchACW Avg ScoreDec 17, 202532
48CursorbenchScoreJul 24, 202631
49PosttrainbenchAverage (%)Jun 16, 202630
50ScienceQAScoreAug 16, 202426
51CL BenchOverallMar 31, 202623
52CybenchUnguided % SolvedFeb 5, 202622
53OSWorldScoreFeb 17, 202620
54ExploitbenchMean capabilityApr 23, 202619
55AlgotuneScoreMar 5, 202618
56CL Bench LifeOverallApr 24, 202617
57Blueprint Bench 2ScoreJul 9, 202617
58Surface Evolver BenchMean scoreJul 16, 202617
59GbaevalOverall scoreJun 16, 202616
60TheAgentCompany% ScoreSep 29, 202516
61CAD EvalOverall pass (%)Apr 16, 202515
62Adversarial NLIScoreApr 23, 202415
63FrontiercodeMain scoreJul 24, 202614
64FrontiersweDominanceJun 9, 202612
65GDPvalWin Rate (%)Dec 11, 202511
66RLIScoreJun 9, 202610
67SuperGLUEScoreJan 27, 202210
68GDP PDFGDP.pdf scoreJun 9, 20269
69OSWorld 2Binary accuracyJun 1, 20269
70Spatialviz BenchOverall scoreJun 17, 20258
71Btf3Pooled scoreJul 9, 20268
72CommonsenseQA 2ScoreJul 18, 20236
73MindcubeOverall scoreMay 22, 20255
SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.
Browse all benchmarks →