Atlas

Compare Models

GPT-5.6 Sol vs GPT-5.6 Terra

GPT-5.6 Sol leads by 2.2 capability points (≈95% interval +0.1 to +4.3): a clear lead over GPT-5.6 Terra · P(higher) ≈ 98%

Wins 281 of 361 shared benchmarks (45 lost, 35 tied)

Capability vs. relative cost
GPT-5.6 SolGPT-5.6 TerraClosed 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-5.6 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins281 – 45 · 361 shared, 35 tied

Comparing matching setups only (18 benchmarks excluded) · suites counted once
← GPT-5.6 SolGPT-5.6 Terra →

GPT-5.6 SolvsGPT-5.6 Terrawins by categoryData Nov 15, 2026Data: Vals AI, OpenAI, Artificial Analysis and 97 othersVillage LLM Atlasatlas.village.dev

Margins

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

SnakeBench Win RateGames100.050.0−50.0
DuetBench–Diagnosis EloHealth & Medicine990952−45.8
NanoGPT - OpenAI self-improvement percentageCode9.714.5+37.5
DecepEval Long-horizon (Induced)Safety47.884.7−36.9
Voratiq Coding Agent LeaderboardCode22991904−35.6
NanoGPT Speedrun Frontier (human record gap closed)Agents35.911.0−33.5
ExploitBench - release-reported exploit progressCybersecurity78.546.2−32.3
TextQuests Average Cumulative HarmSafety31.422.2+31.0
KernelGen1P - OpenAI release percentageCode61.149.2−30.7
PMPP-HardCode59.431.9−27.5
SnakeBench Average ApplesGames18.012.7−26.8
GameBench 2Games80.153.5−26.6
SimpleQA VerifiedGeneral QA69.743.2−26.5
Olam Social Poker EloGames15341511−25.0
Opus Magnum CampaignGames70.947.7−23.1
Mystery Game PuzzlesGames58.035.0−23.0
IOICode86.765.3−21.4
SEC-Bench Pro - release-reported cohortCybersecurity79.157.7−21.4
SRE Bench — Fully SolvedCode30.59.5−21.0
Concept Synth INDUCTION Challenge64 Round 1General QA57.837.5−20.3
Vending-Bench Arena: Round 10Games78219191+20.3
CadQueryEval (Corrected Geometry)Code88.068.0−20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal0.020.0+20.0
CadQueryEvalCode76.056.0−20.0
ChessBench CoherenceGames89.169.7−19.4
LMArena Agent Arena Net ImprovementAgents6.30.6−18.6
Arena Code WebDev — Reference-Based DesignCode16531516−18.1
Concept Synth INDUCTION Challenge100 Round 1General QA43.025.0−18.0
PointerbenchMultimodal65.347.3−18.0
Hedge-BenchFinance & Accounting31.513.8−17.7
ChessBench EloGames1218925−17.4
Odysseys2Agents62.044.6−17.4
HardGeoBench Avg DistanceMultimodal14291991−17.3
Arena Code WebDev — FullstackCode16301565−16.9
APEX-SWECode48.031.1−16.9
DeepsecBenchCode35.619.2−16.4
DeepsecBench — Revalidated V1Code35.419.1−16.3
DeepsecBench — Revalidated V1 (233 issues)Code35.319.1−16.2
SimpleBenchGeneral QA64.848.9−15.9
SpatialViz-BenchMultimodal80.765.2−15.5
Terminal-Bench 4.0Code37.922.7−15.2
Arena Image-to-WebDevCode16061539−14.6
Arena Code WebDev — Content Creation ToolsCode15941503−14.4
Arena Code WebDev — ReactCode16181511−14.3
ScBench-LongScience38.123.8−14.3
Agent Red Teaming k=100 (CAIS)Safety40.354.6−14.2
MindCube TinyMultimodal76.862.6−14.2
SnakeBench RatingGames24.629.8+14.2
AutomationBench 1.0.6 (Zapier)Agents28.814.8−14.0
Arena Code WebDev — GamingCode16721548−14.0
GraphWalks BFS - 256KGeneral QA90.776.9−13.8
Vending-Bench 2Games96197343−13.8
ITSMBenchAgents43.830.1−13.7
Frontier-Bench v0.1Agents34.420.8−13.6
Arena Code WebDev — Brand & MarketingCode16221530−13.3
MedChemBench - OpenAI internal release scoreScience48.335.0−13.3
Arena Code WebDev — FrontendCode16231521−13.1
OSWorld 2.0 500-step Partial ScoreAgents62.650.2−12.4
Arena Code WebDev — HTMLCode16121523−12.4
HackerRank-BenchCode38.426.1−12.3
MRCR v2 (8-needle, Context Arena) - 128K pointwiseGeneral QA92.480.2−12.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents4.6-1.8−12.2
FrontierMath Tier 4 (v2) - PrivateMath82.970.7−12.2
BabaIsBench Stage 0: The IntroGames100.088.0−12.0
Chess Puzzles (task version 2.0)Games59.047.0−12.0
SWEeper-BenchCode47.035.0−12.0
NoCoT-Bench NCKI (kspine v3)General QA121104−11.7
TextQuests Game ProgressGames52.140.5−11.6
Extended NYT Connections 940-Puzzle SetGames96.184.7−11.4
Arena Code WebDev — SimulationsCode16261526−11.3
NoCoT-Bench NCRI 15.2General QA10989.0−11.2
APEX-AccountingFinance & Accounting51.540.3−11.2
Document ArenaProfessional Work14801468−11.1
ChartographyMultimodal45.034.0−11.0
AA-Omniscience IndexIndexes22.00.1−11.0
Bug Hunt BenchCode41.430.5−11.0
HyperBrowseCompSearch26.715.8−10.9
Arena Code WebDev — Data & AnalyticsCode15681496−10.8
DeepSWE v1.1Code70.760.2−10.6
LLM Chess EloGames15501293−10.5
WeirdML (v2)Science88.878.3−10.5
Bio Capabilities Index (SecureBio)Indexes169155−10.3
Gray Swan IPI Attack Success (k=15)Safety27.037.3−10.3
SWE-Marathon v1.1Code42.532.5−10.0
Terminal-Bench Science 0.1 (Vals)Science20.010.0−10.0
Humanity's Last ExamGeneral QA45.535.9−9.6
DecepEval Coding (Neutral)Safety1.611.1−9.5
Surface Evolver BenchAgents93.183.8−9.4
Artificial Analysis Intelligence IndexIndexes60.951.6−9.3
ScBench-Long V0 VerifiedScience43.934.9−9.1
ALE-BenchCode24102154−9.0
ProofBench v1.1Math83.074.0−9.0
SnakeBench Best ApplesGames18.021.0+8.8
DecepEval Tool-use (Induced)Safety36.845.5−8.7
Arena Code WebDev — Consumer ProductCode15731505−8.7
ARC-AGI-2General QA92.583.9−8.6
Terminal-Bench-Science 0.1Science22.413.8−8.6
AA-Briefcase EloAgents14751330−8.4
SWE Atlas - Codebase QnACode53.545.2−8.3
HPCT (SecureBio)Science76.368.2−8.1
RiemannBenchMath74.466.4−8.0
Vals IndexIndexes73.165.1−8.0
ZeroBench Main Mean pass@1 (five samples)Multimodal22.014.2−7.8
MRCR v2 (8-needle, Context Arena) - 128K AUCGeneral QA97.189.4−7.7
ERQAMultimodal71.263.6−7.6
Vals Multimodal IndexIndexes72.665.1−7.6
VCT v2 (SecureBio)Science49.441.9−7.5
GDPval-AA v2 (normalized)Professional Work61.654.1−7.5
CritPt v2 (agentic, avg@5)Science77.169.7−7.4
A-FantasiaGeneral QA36.043.3−7.3
A-Fantasia (Valid Attempts with Retries)Games36.043.3−7.3
Senior SWE-BenchCode34.727.4−7.3
Humanity's Sixth SenseMultimodal30.022.7−7.3
HieroglyphBenchMultimodal27.620.5−7.1
VCT (SecureBio)Science54.347.2−7.1
Diligence Stack Agent BenchFinance & Accounting88.381.3−7.0
Machiavelli Deception (CAIS)Safety82.780.4+7.0
Shroombench (mushroom identification, top-1)Multimodal42.035.0−7.0
BrowseComp Extended (Mercor held-out)Search32.325.3−7.0
ARC-AGI-3 (Semi-Private)General QA7.80.8−7.0
GDPval-AA v2.1Professional Work16191461−6.9
SnakeBench Total ApplesGames36.0127+6.9
Wiz Cyber Model Arena v3.9Cybersecurity69.863.0−6.8
BALROGGames60.053.2−6.8
Legal Research BenchLegal48.141.3−6.7
Artificial Analysis Agentic IndexIndexes54.047.4−6.6
Ramp SWE-Bench (78 published tasks)Code83.376.9−6.4
LAB-Bench CloningScenarios (SecureBio)Science98.892.5−6.3
BioTIER RefusalSafety71.965.6−6.3
Toloka Arena (v1)Agents47.641.4−6.2
Excel Modeling BenchmarkFinance & Accounting72.366.2−6.1
SUPERChemScience71.064.9−6.1
BullshitBench v2 LeaderboardSafety48.054.0+6.0
GDP.pdfProfessional Work30.724.7−6.0
OpenAI management-consulting tasks - internal release scoreProfessional Work43.237.2−6.0
ProofBenchMath77.071.0−6.0
WordleBenchGames97.091.0−6.0
Creative Writing v3 EloChat & Writing19741856−5.9
Vibe Code Bench v1.1Code80.574.6−5.9
GraphWalks BFS - 1MGeneral QA77.171.2−5.9
SAGEMath52.647.0−5.6
Medical Long Context Reasoning (MLCR-AA)Health & Medicine26.131.7+5.6
AHK-EvalCode94.488.9−5.6
Vision Arena OverallMultimodal12881266−5.5
SkillsBench without SkillsAgents43.348.8+5.5
GeneBench-ProScience28.723.3−5.4
LMArena Agent Arena Bash ReliabilityAgents11.40.7−5.3
PolicyBench US Exact MatchFinance & Accounting88.783.4−5.2
GDPval-AA v2Professional Work15851479−5.2
BenchCAD - geometric overlapMultimodal83.478.2−5.2
BioSecBench-RefusalSafety39.534.3−5.2
ITBench-AAAgents56.251.0−5.2
Code MigrationCode52.947.8−5.1
GPQA DiamondGeneral QA94.989.9−5.0
AA-LCR v1.1Search84.079.0−5.0
HardGeoBench <=100miMultimodal41.336.3−5.0
HardGeoBench <=10miMultimodal30.025.0−5.0
MMLU Pro - EngineeringGeneral QA87.882.9−5.0
Vals Index v2.1Indexes58.053.1−4.9
Frontier Financial JudgementFinance & Accounting51.246.3−4.9
LAB-Bench ProtocolQA (SecureBio)Science81.977.0−4.9
OpenAI Capture-the-Flag challenges - internal release scoreCybersecurity96.791.8−4.9
ToolathlonAgents58.053.1−4.9
FrontierCode 1.1 Extended - Pass RateCode66.661.8−4.9
ChessBench AccuracyGames76.471.9−4.8
BioSecBench-Function V0Science38.333.5−4.8
SkillsBench with SkillsAgents54.158.9+4.8
LMArena Agent Arena Confirmed SuccessAgents2.3-7.2−4.8
MMLU Pro - LawGeneral QA79.174.4−4.7
FrontierCode 1.1 Extended - ScoreCode60.555.8−4.7
WCB (SecureBio)Science64.459.8−4.6
SAB RankingForecasting-0.1-0.0+4.5
Terminal-Bench 4.0 (Artificial Analysis)Code39.935.4−4.5
CVDP (517-problem set)Code55.450.9−4.5
PCT Political Inconsistency (CAIS)Safety41.746.2−4.5
UnityBench-Mini3DCode72.568.0−4.5
EnterpriseOps-Gym-AAAgents42.938.5−4.4
MMLU Pro - PhilosophyGeneral QA88.484.0−4.4
DecepEval Coding (Induced)Safety31.727.3+4.4
GBENCH One-shot Coding GScoreGames67.763.3−4.3
LisanBench Path LengthGames46993303−4.3
DeepResearch Bench II (Mercor 130-task set)Search51.547.2−4.3
RCM-Bench-1Health & Medicine91.287.0−4.3
CyberBench v1.1 OverallCybersecurity76.372.1−4.2
MBCT (SecureBio)Science61.156.9−4.2
Conceptual Reasoning IndexIndexes68.765.7−4.2
EQ-Bench 4Chat & Writing12361220−4.2
LMArena Agent Arena Praise vs ComplaintAgents6.3-1.9−4.1
Public Benefits Bench v1.1Professional Work66.562.4−4.1
𝜏³-BankingAgents44.340.2−4.1
Vals Index v2Indexes63.759.6−4.1
SpatialBench V0 VerifiedScience74.570.4−4.1
DecepEval Long-horizon (Neutral)Safety1.55.5−4.0
τ^τ-bench release v1 (hyper-tau-bench)Agents22.018.0−4.0
Financial Coaching BenchFinance & Accounting74.971.0−3.9
LifeSciBench - release-reported version unspecifiedScience59.956.0−3.9
Longform Creative WritingChat & Writing81.778.0−3.7
DAYJOB: FinanceFinance & Accounting5.51.8−3.7
Finance Agent v2 - Disclosure AnalysisFinance & Accounting63.366.9+3.6
MMLU Pro - OtherGeneral QA87.784.1−3.6
IOI (Vals v2)Code91.287.6−3.6
Finance Agent v2 - Market AnalysisFinance & Accounting64.260.6−3.5
LMCAGeneral QA58.555.0−3.5
FrontierPhysics v0.1Science7.13.6−3.5
VariantBenchScience42.138.7−3.4
LiveCodeBench (Vals AI public dataset)Code82.685.9+3.3
YC-BenchGames728001636144−3.2
HealthBench Hard (updated implementation)Health & Medicine31.134.3+3.2
FrontierMath Tiers 1-3 (v2) - PrivateMath89.186.0−3.2
SWE-sweepCode4.71.5−3.1
SWE-Bench Pro V2 — FullCode95.592.4−3.1
HealthBench (updated implementation)Health & Medicine55.658.7+3.1
LMArena Agent Arena Steering BurdenAgents5.5-0.5−3.0
SciCode (Artificial Analysis avg@3)Science56.953.9−3.0
AA-LCRSearch71.074.0+3.0
Eval Connections One-Shot with TrapsGames83.086.0+3.0
LLM Sycophancy (Balanced 198)Safety3.00.0+3.0
ZeroBench Main Questions pass^5Multimodal13.010.0−3.0
LMArena Text Arena EloChat & Writing14851468−3.0
Agents' Last Exam (ALE)Agents30.627.6−3.0
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting75.672.7−2.9
Finance Agent v2 - Earnings AnalysisFinance & Accounting74.477.3+2.9
BrowseCompSearch90.487.5−2.9
AutomationBench (Overall)Agents18.115.2−2.9
MMLU Pro - PsychologyGeneral QA89.286.3−2.9
DistressBenchSafety73.676.4+2.9
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting64.967.8+2.9
Blueprint-Bench 2Multimodal0.30.3−2.8
TxBench Antibody Discovery V0Science33.831.0−2.8
Tax Agent BenchFinance & Accounting68.065.2−2.8
Artificial Analysis Progress Index v1Indexes331324−2.7
IntPhys 2Multimodal63.961.3−2.6
MMMU ProMultimodal84.682.0−2.6
Furniture Assembly BenchmarkMultimodal56.754.2−2.5
HardGeoBench Country AccuracyMultimodal47.545.0−2.5
RedlineBenchLegal50.347.8−2.5
MMLU Pro (Vals 5-Shot CoT)General QA89.186.7−2.4
MMLU Pro - BusinessGeneral QA93.390.9−2.4
DTBenchGeneral QA95.793.3−2.4
MedScribeHealth & Medicine85.282.9−2.4
MMLU Pro - HistoryGeneral QA81.979.5−2.4
Finance Agent v2 (Vals Index subset)Finance & Accounting55.052.6−2.3
Finance Agent v2 (Vals Multimodal Index component)Finance & Accounting55.052.6−2.3
Finance Agent v2 - Financial ModelingFinance & Accounting20.723.0+2.3
GeneBench - release-reported version unspecifiedScience30.728.4−2.3
CursorBench 3.2Code67.264.9−2.3
DecepEval Tool-use (Neutral)Safety2.24.5−2.3
LisanBench Difficulty-WeightedGames14031197−2.2
EpiBenchScience42.139.9−2.2
MASKSafety95.493.3−2.1
SM BenchSafety61.159.0−2.1
Roboflow Vision EvalsMultimodal78.376.2−2.1
Big Finance Bench - OpenAI release scoreFinance & Accounting53.051.0−2.0
ChessBench — Historical Output RepairGames42.740.7−2.0
CRED v0.92Science99.097.0−2.0
OpenAI MRCR v2 8-needle - 256K-512KGeneral QA91.589.6−1.9
VulcanBench Frontier v4Code87.289.1+1.9
LegalBenchLegal87.085.1−1.9
MMLU Pro - ChemistryGeneral QA92.090.2−1.8
Finance Agent v2 - ComparablesFinance & Accounting45.347.0+1.7
Epoch Capabilities Index (ECI)Indexes163161−1.7
LAB-Bench LitQA2 (SecureBio)Science52.851.1−1.7
HealthBench ProfessionalHealth & Medicine64.162.4−1.7
MMLU Pro - PhysicsGeneral QA92.891.1−1.7
Artificial Analysis Coding IndexIndexes78.376.7−1.7
MMLU Pro - EconomicsGeneral QA90.689.0−1.7
CorpFin v2 - Exact PagesFinance & Accounting61.563.2+1.6
RSI Index - OpenAI internal release compositeIndexes57.956.3−1.6
Humanity's Last Exam DiamondGeneral QA30.428.8−1.6
MMLU Pro - BiologyGeneral QA94.693.0−1.5
SEA-HELMGeneral QA77.075.4−1.5
SpatialBench-Long V0 VerifiedScience36.437.9+1.5
Terminal-Bench HardCode61.462.9+1.5
BioSecBench-SurveillanceScience52.551.0−1.5
IFBench (Artificial Analysis avg@5)Chat & Writing72.771.2−1.4
CritPt (avg@5)Science28.630.0+1.4
Finance Agent v2 - All PassFinance & Accounting41.442.8+1.4
TaxEval v2Finance & Accounting74.876.2+1.4
ObviousBenchGeneral QA96.597.9+1.4
Creative Writing v3 Rubric ScoreChat & Writing16.215.9−1.4
Finance Agent v2 - PrecedentsFinance & Accounting26.224.8−1.3
MMLU Pro - HealthGeneral QA84.182.8−1.3
CharXiv ReasoningMultimodal84.685.9+1.3
OpenAI MRCR v2 8-needle - 512K-1MGeneral QA73.872.5−1.3
LiveBench 2.0Agents76.375.0−1.3
ACCoRDGeneral QA75.974.7−1.2
LMArena Agent Arena SteerabilityAgents6.54.1−1.2
PostTrainBench Lite - release-reported performanceCode50.351.5+1.2
TxBench Oligo Discovery V0Science34.533.3−1.2
Ruby LLM Benchmarks — Program FixerCode75.274.0−1.2
τ²-bench (Telecom)Agents85.186.3+1.2
CorpFin v2 - Max Fitting ContextFinance & Accounting65.266.3+1.2
MMLU Pro - MathMath95.394.2−1.1
ARC-AGI-1 Semi-PrivateGeneral QA97.596.5−1.0
HPCT Refusal (CAIS)Safety24.025.0+1.0
ProgramBenchCode1.50.5−1.0
Terminal-Bench 2.1 Extended (Mercor held-out)Code37.436.4−1.0
SpeciEval (Pooled English Runs)Safety99.098.1−1.0
SpeciEvalSafety99.098.1−1.0
SpeciEval (Refusal-Aware Retries, Historical)Safety99.098.1−1.0
Political Science LLM BenchmarkProfessional Work69.268.3−1.0
DiligenceBenchFinance & Accounting50.049.1−0.9
SWE-bench VerifiedCode96.295.4−0.8
Harvey HLAB — Criteria Pass RateLegal80.579.8−0.8
ReactBench v1Code52.553.3+0.8
PubMedQA (SecureBio)Health & Medicine81.180.4−0.7
VCT-Refusal (CAIS)Safety11.310.7−0.6
MedCodeHealth & Medicine44.043.4−0.6
MentalHealthBenchSafety46.746.2−0.5
OpenAI research debugging - internal release scoreCode68.367.8−0.5
Finance Agent v2 - AdjustmentsFinance & Accounting49.249.7+0.5
CursorBench 4.0Code41.741.3−0.4
HealthBench Hard (length-adjusted)Health & Medicine33.132.7−0.4
SciCode (main problems)Science16.215.8−0.4
MMLU-Bio (SecureBio)Science91.190.7−0.4
CRM-Agent-BenchAgents81.381.6+0.4
BuseyBench SVGMultimodal6.16.1−0.3
OTIS Mock AIME 2024-2025Math100.099.7−0.3
SAB ResearchFinance & Accounting93.693.4−0.3
MMLU Pro - Computer ScienceGeneral QA90.791.0+0.2
RuneBenchGames5.95.9−0.2
HealthBench Consensus (updated implementation)Health & Medicine95.395.2−0.1
LMArena Agent Arena Bash RecoveryAgents0.70.9+0.1
MortgageTax (Vals all 1,258 documents)Finance & Accounting67.367.30.0
LMArena Agent Arena Tool HallucinationAgents1.01.00.0
Aikido Known-CVE Detection (Pass@3 Union)Cybersecurity88.588.50.0
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science5.85.80.0
Continual Learning Exam (Trial 1)Agents27.827.80.0
CorpFin v2 - Shared Max ContextFinance & Accounting66.466.40.0
CorpFin v2 (Vals Index component)Finance & Accounting66.466.40.0
CorpFin v2 (Vals Multimodal Index component)Finance & Accounting66.466.40.0
DAYJOB: HealthcareHealth & Medicine1.61.60.0
Eval ConnectionsGames100.0100.00.0
GPQA-Bio (SecureBio)Science85.385.30.0
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 - Corporate M&ALegal0.00.00.0
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 - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
HealthBench (length-adjusted)Health & Medicine57.057.00.0
Terminal-Bench 2.1Code88.088.00.0
Visual Acuity 4K Threshold RowMultimodal12.012.00.0
WMDP-Bio (SecureBio)Science89.789.70.0

GPT-5.6 SolvsGPT-5.6 Terranormalised score marginsData Nov 15, 2026Data: Vals AI, OpenAI, Artificial Analysis and 97 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…