Atlas

Compare Models

GPT-5.6 Sol vs GPT-6 Astra

GPT-6 Astra leads by 4.4 capability points (≈95% interval +1.9 to +6.8): a clear lead over GPT-5.6 Sol · P(higher) ≈ 100%

Wins 289 of 364 shared benchmarks (49 lost, 26 tied)

Capability vs. relative cost
GPT-5.6 SolGPT-6 AstraClosed weightsOpen weights

Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting.

GPT-5.6 SolvsGPT-6 Astracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins49 – 289 · 364 shared, 26 tied

Comparing matching setups only (28 benchmarks excluded) · suites counted once
← GPT-5.6 SolGPT-6 Astra →

GPT-5.6 SolvsGPT-6 Astrawins by categoryData Nov 15, 2026Data: Vals AI, benchmarks.bio, OpenAI and 110 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6 Astra minus GPT-5.6 Sol.

Time Horizon Index: KSPAgents23.890.5+66.7
OpenScore String Quartets - 1 minus OMR-NEDMultimodal0.20.8+65.0
ARC-AGI-3 (Semi-Private)General QA7.862.7+54.9
ActiveVisionMultimodal23.576.5+53.0
Concept Synth INDUCTION Challenge100 Round 1General QA43.094.0+51.0
Design Creativity BenchCode51.365.5+45.0
TasteVal (compute multiplier)Science0.61.6+44.4
Clocktower BenchGames15831892+44.3
Terminal-Bench Science 0.1 (Vals)Science20.062.9+42.9
Terminal-Bench-Science 0.1Science22.464.6+42.2
Concept Synth INDUCTION Challenge64 Round 1General QA57.898.4+40.6
Short Form Bench v1.0Games24102803+38.9
SnakeBench RatingGames24.637.9+35.9
CyberBench v1.1 OverallCybersecurity76.341.1−35.2
Terminal-Bench 4.0 (Artificial Analysis)Code24.759.6+34.8
ExploitBench June-August 2026 - ACECybersecurity5.539.0+33.5
Humanity's Last Exam DiamondGeneral QA30.463.8+33.4
HiL-Bench (Human-in-Loop Benchmark)Agents32.365.3+33.0
FrontierSWE V2 Worst@5Code22.255.1+32.9
SRE-Bench - OpenAI release pass@1Code55.988.0+32.1
ChessBench EloGames12181755+31.9
BrokenArXiv 06/2026Math67.399.1+31.8
EBR-bench v4 (card ban, single agent)Games44.876.2+31.4
CheatBenchSafety74.945.0+29.9
FrontierSWE V2 Best@5Code44.273.0+28.8
ExploitBench - Internal Port (June-August 2026)Cybersecurity3.531.5+28.0
ALE-BenchCode23913164+27.4
NoCoT-Bench NCRI 15.2General QA109159+27.2
WeirdML v3Science15.042.2+27.2
Agent Red Teaming k=100 (CAIS)Safety40.313.5+26.8
MirrorCode (ML, +Private, 2L)Code20.046.7+26.7
SnakeBench Best ApplesGames18.027.0+26.5
SRE Bench — Fully SolvedCode30.556.9+26.3
ArXivLean 06/2026Math39.165.2+26.1
Mystery Game PuzzlesGames58.084.0+26.0
ChartographyMultimodal45.071.0+26.0
HPCT Refusal (CAIS)Safety24.050.0+26.0
GlobeBenchGeneral QA23.749.6+25.9
FrontierCUA pass@1Agents16.442.2+25.8
Biopharma Bench V0.1Health & Medicine44.970.4+25.5
Arena Image-to-WebDevCode16061722+25.2
Vals Science IndexIndexes41.565.1+23.6
Humanity's Sixth SenseMultimodal30.053.6+23.6
AutoformBenchMath58.982.4+23.5
VIALSMultimodal26.549.9+23.4
VoxelBench Text RatingGames22972625+23.4
Furniture Assembly BenchmarkMultimodal56.780.0+23.3
ChessBench — Historical Output RepairGames42.766.0+23.3
AutomationBench (Overall)Agents18.141.4+23.3
TxBench Antibody Discovery V0Science33.856.9+23.1
SciUniverse Level 1Science9.432.5+23.0
Odysseys2Agents62.084.6+22.6
OpenAI MRCR v2 8-needle - 512K-1MGeneral QA73.896.3+22.5
TextQuests Game ProgressGames52.174.2+22.1
ABC-Bench liquid handling (SecureBio)Science98.776.7−22.0
CrossSearch (pass^3)Search38.060.0+22.0
ZeroBench Main Questions pass^5Multimodal13.035.0+22.0
Terminal-Bench 4.0Code37.959.6+21.7
Search ArenaSearch12581231−21.7
ExploitBench - release-reported exploit progressCybersecurity78.5100.0+21.5
SCI-AGENT-EXAMScience59.680.9+21.3
ZeroBench Main Mean pass@1 (five samples)Multimodal22.043.2+21.2
OpenAI internal database migration - reported scoreCode42.763.9+21.2
BullshitBench v2 LeaderboardSafety48.069.0+21.0
TxBench Oligo Discovery V0Science34.555.5+21.0
RuneBenchGames5.97.3+20.9
ORCA-bench Hard RCA (Fireworks 324-task subset)Agents29.550.2+20.7
Harvey's Legal Agent Benchmark - Corporate M&ALegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal0.020.0+20.0
RareDiseaseBench V1 (35 evaluations)Science13.333.3+20.0
Real-SWECode26.346.3+20.0
SnakeBench Win RateGames100.080.0−20.0
MysteryMechanismScience33.353.2+19.8
OpenAI internal computer-use safety - adverse outcome rateSafety22.02.4+19.6
LMArena Code Arena WebDevCode16151764+19.5
Machiavelli Deception (CAIS)Safety82.776.3+19.3
Learn2Play BenchGames57.076.2+19.2
Drone-BenchAgents76.195.1+19.0
LMArena Agent Arena Net Improvement (Steering Burden)Agents4.614.5+18.7
Gray Swan IPI Attack Success (k=15)Safety27.08.5+18.5
MetagenomicsBench V0Science31.349.7+18.4
MetagenomicsBench V1Science31.349.7+18.3
LisanBench Difficulty-WeightedGames14033113+18.3
TissueMicroscopyBench V1 (49 evaluations)Science2.019.7+17.7
TxBench Antibody Discovery V1 Clean RerunsScience31.448.9+17.5
Olam Social Poker EloGames15341550+17.4
Blueprint-Bench 2Multimodal0.30.5+16.1
FrontierPhysics v0.1Science7.123.2+16.1
HardGeoBench Avg DistanceMultimodal1429908+16.1
DAYJOB: FinanceFinance & Accounting5.521.5+16.0
ProofBench v1.1Math83.099.0+16.0
MindCube TinyMultimodal76.892.7+15.9
ScreenSpot-ProMultimodal76.992.7+15.8
FrontierCode 2.0 - Pass RateCode32.347.6+15.3
FrontierCode 2.0 - ScoreCode28.944.0+15.1
Chess Puzzles (task version 2.0)Games59.074.0+15.0
Code MigrationCode52.967.7+14.8
FrontierMath Tier 4 (v2) - PrivateMath83.097.6+14.6
GameBench 2Games80.194.3+14.3
TxBench Oligo Discovery V1 Clean RerunsScience35.049.2+14.2
TaxCalcBench TY25Finance & Accounting30.044.0+14.0
TxBench-Antibody-Development V1 (60 evaluations)Science18.932.8+13.9
LMArena Agent Arena Net ImprovementAgents6.310.4+13.5
IntPhys 2Multimodal63.977.4+13.5
SWE Atlas - Codebase QnACode46.059.1+13.1
BioSecBench-Function V1 Clean RerunsScience35.548.6+13.1
Hedge-BenchFinance & Accounting31.544.2+12.7
LMArena Agent Arena Praise vs ComplaintAgents6.331.7+12.7
AutomationBench 1.0.6 (Zapier)Agents28.841.4+12.6
RedlineBenchLegal50.362.8+12.6
BenchCAD - geometric overlapMultimodal83.495.9+12.5
HardGeoBench Country AccuracyMultimodal47.560.0+12.5
SpatialViz-BenchMultimodal80.793.1+12.4
HieroglyphBenchMultimodal27.640.0+12.4
SnakeBench Total ApplesGames36.0198+12.3
Agents' Last Exam v1.1 (ALE v1.1)Agents30.342.6+12.3
SWE-Shift (Shifted)Code12.624.8+12.2
BlenderWorldCode67.779.9+12.2
BuseyBench SVGMultimodal6.17.3+12.1
ExploitGym - release-reported percentage, cohort unspecifiedCybersecurity30.342.4+12.1
MentalHealthBenchSafety46.758.7+12.0
CadQueryEval (Corrected Geometry)Code88.0100.0+12.0
GeneBench-Pro Public Release SubsetScience24.036.0+12.0
Shroombench (mushroom identification, top-1)Multimodal42.054.0+12.0
GBENCH One-shot Coding GScoreGames67.779.2+11.6
Document ArenaProfessional Work14801467−11.4
VariantBench V1 Clean RerunsScience35.346.6+11.3
ERQAMultimodal71.282.3+11.1
LLM Creative Story WritingChat & Writing2.53.4+11.0
Eval Connections One-Shot with TrapsGames83.094.0+11.0
Roboflow Vision Evals (October 2026, five tasks)Multimodal72.483.4+11.0
OpenDS-WorldSearch28.039.0+11.0
AA-Omniscience IndexIndexes22.043.7+10.9
PostTrainBench v1.2Indexes31.041.9+10.9
CUA-benchAgents8.319.2+10.8
Dual Exam Agent IndexIndexes20.731.3+10.6
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting44.254.6+10.4
BioSecBench-Function V0Science38.348.6+10.3
Humanity's Last ExamGeneral QA44.554.8+10.3
Creative Writing v3 EloChat & Writing19742177+10.3
SWE-Clar (implementation)Code57.067.0+10.0
DAYJOB: HealthcareHealth & Medicine1.611.6+10.0
HardGeoBench <=100miMultimodal41.351.3+10.0
LiveBench 2.0Agents76.386.1+9.9
FrontierSWE V2 (Mercor rerun, mean@3)Code20.229.9+9.7
SWEeper-BenchCode47.056.5+9.5
SWE-Clar (clarification)Code78.187.5+9.4
EurekaBenchScience19.428.7+9.3
SUPERChemScience71.080.3+9.3
ChessBench AccuracyGames76.485.0+9.2
SnakeBench Average ApplesGames18.019.8+9.1
Into the Unknown Integrity Node Pass RateScience88.797.8+9.0
DrugDiscoveryBenchScience56.565.4+8.9
Medical Long Context Reasoning (MLCR-AA)Health & Medicine26.135.0+8.9
IOI (Vals v2)Code91.2100.0+8.8
Co-Design BenchCode61.169.9+8.8
SM BenchSafety61.169.9+8.8
Into the Unknown Direction Node Pass RateScience74.983.6+8.7
AutoResearchExam (hidden-test AUARC)Agents0.50.6+8.7
Legal Research BenchLegal48.139.4−8.7
VCT v2 (SecureBio)Science49.458.0+8.6
OpenAI MRCR v2 8-needle - 256K-512KGeneral QA91.5100.0+8.5
Design Arena Overall Code CategoriesCode13381389+8.5
AutomationBench-AA 1.0.6Agents60.168.5+8.4
LisanBench Path LengthGames46997393+8.4
BALROGGames60.068.3+8.3
BioMysteryBench (Vals Terminus 2)Science71.179.3+8.1
BioTIER RefusalSafety71.979.9+8.0
RT-SAFE Collisions per EpisodeSafety25.622.4+8.0
FORTRESS Average Risk Score (ARS)Safety20.112.1+8.0
RSI-Exam 0.1 (Full 88)Agents0.40.5+7.9
ChessBench CoherenceGames89.197.0+7.9
Vibe Code Bench 1–100Code20.027.6+7.6
ITBench-AAAgents56.248.6−7.6
Roboflow Vision Evals (September 2026 protocol)Multimodal79.086.6+7.6
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science5.813.3+7.6
Finance Agent v2 - ComparablesFinance & Accounting45.337.8−7.5
HardGeoBench <=10miMultimodal30.037.5+7.5
LAB-Bench ProtocolQA (SecureBio)Science81.989.3+7.4
Financial Coaching BenchFinance & Accounting74.967.5−7.4
TextQuests Average Cumulative HarmSafety31.429.2+7.3
Nimble Web ContextSearch80.072.6−7.3
HPCT (SecureBio)Science76.369.0−7.3
NoCoT-Bench NCKI (kspine v3)General QA121131+7.3
ComplexConstraintsChat & Writing50.557.7+7.2
Conceptual Reasoning IndexIndexes68.773.9+7.1
Artificial Analysis Coding Agent Index v1.5Indexes54.661.6+7.1
APEX-Agents 1.1 Corporate Law (68-task image-free slice)Legal80.687.7+7.1
WellnessBenchSafety57.064.0+7.0
NeuroscienceBench V1 (91 evaluations)Science15.822.7+7.0
Harvey HLAB — Criteria Pass RateLegal80.587.5+6.9
OSWorld 2 (Aug 8) offline - reported partial reward, step limit unspecifiedAgents65.872.6+6.8
ClockBench AccuracyMultimodal68.375.0+6.7
ATLAS-Finance v1.0Finance & Accounting4.711.3+6.7
VariantBenchScience42.148.6+6.5
APEX-AccountingFinance & Accounting51.557.9+6.4
Artificial Analysis Progress Index v1Indexes331350+6.4
VCT-Refusal (CAIS)Safety11.317.6+6.3
SEC-Bench Pro - release-reported cohortCybersecurity79.185.4+6.3
Toloka Arena (v1)Agents47.653.8+6.2
SAGEMath52.646.4−6.2
CritPt v2 (agentic, avg@5)Science77.183.1+6.0
Toloka Arena (v3)Agents61.067.0+6.0
Vals RSI Index v1.1 (historical five tasks)Agents23.129.0+5.9
ArXivMath 06/2026Math88.594.4+5.9
LMCAGeneral QA58.564.4+5.9
LMArena Agent Arena Confirmed SuccessAgents2.314.0+5.9
ACCoRDGeneral QA75.981.7+5.8
GameDevBenchGames63.168.8+5.7
ATLAS-Finance v1.1Finance & Accounting4.09.7+5.7
LLM Debate BenchmarkChat & Writing16451606−5.6
LibraryDesignBenchCode39.545.1+5.6
AHK-EvalCode94.4100.0+5.6
BioSecBench-RefusalSafety39.544.6+5.1
HackerRank-BenchCode38.443.5+5.1
AA-Briefcase EloAgents14751562+5.1
Epoch Capabilities Index (ECI)Indexes163169+5.0
Finance Agent v2 - Earnings AnalysisFinance & Accounting74.469.5−4.9
SWE Atlas - Test WritingCode45.950.7+4.8
GeneBench Pro v13Science32.337.1+4.8
LMArena Agent Arena Steering BurdenAgents5.514.8+4.6
Tax Agent BenchFinance & Accounting68.063.3−4.6
WitnessBench v1.0Games23.027.6+4.6
Harvey LAB held-out set (Fireworks run)Legal0.65.1+4.6
FrontierMath Tiers 1-3 (v2) - PrivateMath89.193.7+4.6
WeirdML (v2)Science88.893.3+4.5
MedCodeHealth & Medicine44.048.5+4.5
GDP.pdf (Artificial Analysis, All-pass)Professional Work27.832.2+4.4
SpatialBench V0 VerifiedScience74.570.1−4.3
PCT Political Inconsistency (CAIS)Safety41.737.4+4.3
Extended NYT Connections 940 (Quadratic v1)Games93.898.1+4.3
VCT (SecureBio)Science54.358.5+4.2
HealthBench ProfessionalHealth & Medicine64.168.2+4.1
ScBench V1 Clean RerunsScience59.863.9+4.1
Finance Agent v2 - PrecedentsFinance & Accounting26.230.3+4.1
FrontierCode 1.1 Extended - Pass RateCode66.670.6+4.0
AA-LCR v1.1Search84.080.0−4.0
ProgramBenchCode1.55.5+4.0
LMArena Agent Arena SteerabilityAgents6.5-1.4−4.0
FrontierCode 1.1 Extended - ScoreCode60.564.5+3.9
MMMU ProMultimodal82.786.4+3.8
AA-AnalystAgent (pass^5)Professional Work47.551.3+3.7
Vectara Hallucination LeaderboardSafety12.48.7+3.7
Arena Alignment IndexSafety84.287.8+3.6
EnterpriseBench: CoreCraftProfessional Work46.242.6−3.6
HealthBench Hard (length-adjusted)Health & Medicine33.136.6+3.5
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting64.968.4+3.5
ObviousBenchGeneral QA96.5100.0+3.5
SciCode (main problems)Science16.219.6+3.4
DeepSWE v1.1Code70.774.1+3.4
Harvey LAB: In-House Contracting (held-out)Legal2.05.3+3.3
Bio Capabilities Index (SecureBio)Indexes169164−3.3
LABBench2Science82.185.4+3.3
Vals RSI Index v1.1 (four tasks)Agents23.927.1+3.2
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting75.678.8+3.2
CritPt (avg@5)Science28.631.7+3.1
HealthBench Hard (updated implementation)Health & Medicine31.134.2+3.1
MASKSafety95.498.5+3.1
FrontierMath ErdősMath0.02.9+2.9
GSO Opt@1Code76.579.4+2.9
Vals Index v2Indexes63.766.6+2.9
MaxxBenchCode69.171.9+2.8
SWE-TogetherCode57.860.6+2.8
MBCT (SecureBio)Science61.163.9+2.8
ErdosBenchMath3.13.2+2.8
MedScribeHealth & Medicine85.287.9+2.7
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine18.721.3+2.7
LLM Chess EloGames15501614+2.6
EpiBench V1 Clean RerunsScience25.828.3+2.5
EpiBenchScience42.144.6+2.5
ARC-AGI-2General QA92.595.0+2.5
SpeciEval (Pooled English Runs)Safety99.096.5−2.5
SpeciEval (Refusal-Aware Retries, Historical)Safety99.096.5−2.5
SAB RankingForecasting-0.1-0.0+2.4
LMArena Agent Arena Bash RecoveryAgents0.75.5+2.4
RiemannBenchMath74.472.0−2.4
DeepsecBench — Revalidated V1Code35.437.8+2.4
DeepsecBench — Revalidated V1 (233 issues)Code35.337.6+2.3
Bedside BenchHealth & Medicine91.193.4+2.3
ScBench-Long V1 Clean RerunsScience29.627.3−2.3
VulcanBench Frontier v4Code87.289.3+2.1
GPQA-Bio (SecureBio)Science85.387.4+2.1
Senior SWE-BenchCode34.736.8+2.1
TxBench Preclinical Pharmacology V1 Clean RerunsScience61.359.3−2.0
Long-Horizon Terminal-Bench 2.0 (mean reward)Code67.269.2+2.0
LAB-Bench LitQA2 (SecureBio)Science52.854.8+2.0
LMArena Text Arena EloChat & Writing14851474−1.9
WCB (SecureBio)Science64.462.6−1.8
SpatialBench-Long V1 Clean RerunsScience37.539.3+1.8
LMArena Agent Arena Bash ReliabilityAgents11.47.8−1.8
Finance Agent v2 - Disclosure AnalysisFinance & Accounting63.361.7−1.6
Agents' Last Exam (ALE)Agents30.632.2+1.6
HealthBenchHealth & Medicine55.356.9+1.6
GPQA DiamondGeneral QA94.696.2+1.6
DTBenchGeneral QA95.797.3+1.6
SAB ResearchFinance & Accounting93.695.2+1.5
ScBench-Long V0 VerifiedScience43.942.4−1.5
SpatialBench-Long V0 VerifiedScience36.434.9−1.5
ForecastBench Preliminary Dataset Brier IndexForecasting61.462.9+1.5
GDPval-AA v2Professional Work15851555−1.5
Bug Hunt BenchCode41.442.9+1.4
SWE-Bench Pro V2 — FullCode95.596.9+1.4
DeepResearch Bench II (Mercor 130-task set)Search51.550.1−1.4
GDPval-AA v2.1Professional Work16191587−1.4
Finance Agent v2 - Market AnalysisFinance & Accounting64.265.5+1.4
SciCode 1.0.1 (Artificial Analysis)Code57.856.5−1.3
BioSecBench-Surveillance V1 Clean RerunsScience47.348.5+1.3
𝜏³-BankingAgents44.343.1−1.2
LAB-Bench CloningScenarios (SecureBio)Science98.8100.0+1.2
Vision Arena OverallMultimodal12881283−1.2
Creative Writing v3 Rubric ScoreChat & Writing16.215.9−1.2
Finance Agent v2 - All PassFinance & Accounting41.440.3−1.1
BrowseCompSearch90.491.5+1.1
Longform Creative WritingChat & Writing81.782.8+1.1
WMDP-Bio (SecureBio)Science89.790.8+1.1
ARC-AGI-1 Semi-PrivateGeneral QA97.598.5+1.0
MedChemBench - OpenAI internal release scoreScience48.349.3+1.0
Finance Agent v2 - AdjustmentsFinance & Accounting49.248.3−0.9
Finance Agent v2 - Financial ModelingFinance & Accounting20.721.6+0.9
ForecastBench Tournament Overall Brier IndexForecasting66.166.9+0.8
DistressBenchSafety73.674.3+0.7
Excel Modeling BenchmarkFinance & Accounting72.371.7−0.6
APEX-SWECode48.048.6+0.6
Wiz Cyber Model Arena v3.9Cybersecurity69.869.2−0.6
SpatialBench V1 Clean RerunsScience68.167.5−0.6
BioSecBench-Structure V1 (58 evaluations)Science20.721.3+0.6
CRED v0.92Science99.099.5+0.5
CVDP (517-problem set)Code55.454.9−0.5
Terminal-Bench 2.1Code89.589.1−0.4
Artificial Analysis Intelligence IndexIndexes60.961.2+0.3
Bench to the Future 3 — September Panel — PooledForecasting0.10.1+0.2
MacroscopeBenchCode77.878.0+0.2
MMLU-Bio (SecureBio)Science91.191.3+0.2
Prophet Arena Arena Score (Fixed Context)Forecasting84.884.7−0.1
PubMedQA (SecureBio)Health & Medicine81.181.2+0.1
SEA-HELMGeneral QA77.076.9−0.1
Ruby LLM Benchmarks — Program FixerCode75.275.10.0
LMArena Agent Arena Tool HallucinationAgents1.01.00.0
BabaIsBench Stage 0: The IntroGames100.0100.00.0
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Arbitration International Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Banking FinanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Bankruptcy RestructuringLegal0.00.00.0
Harvey's Legal Agent Benchmark - Capital MarketsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal0.00.00.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal0.00.00.0
Harvey's Legal Agent Benchmark - Employment LaborLegal0.00.00.0
Harvey's Legal Agent Benchmark - Environmental ESGLegal0.00.00.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.00.00.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal0.00.00.0
Harvey's Legal Agent Benchmark - ImmigrationLegal0.00.00.0
Harvey's Legal Agent Benchmark - InsuranceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Intellectual PropertyLegal0.00.00.0
Harvey's Legal Agent Benchmark - International Trade SanctionsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Litigation Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Real EstateLegal0.00.00.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal0.00.00.0
Harvey's Legal Agent Benchmark - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
Into the UnknownScience0.00.00.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
SWE-Marathon v1.1Code42.542.50.0

GPT-5.6 SolvsGPT-6 Astranormalised score marginsData Nov 15, 2026Data: Vals AI, benchmarks.bio, OpenAI and 110 othersVillage LLM Atlasatlas.village.dev

Win counts use each benchmark’s declared score direction. Represented suite slices and the derived Capability score are excluded from headline totals. Missing or incompatible setups are excluded, never counted as losses.
Loading Atlas data…