Atlas

Compare Models

GPT-5.6 Terra vs GPT-6 Sol

GPT-6 Sol leads by 3.1 capability points (≈95% interval +0.9 to +5.2): a clear lead over GPT-5.6 Terra · P(higher) ≈ 100%

Wins 160 of 227 shared benchmarks (42 lost, 25 tied)

Capability vs. relative cost
GPT-5.6 TerraGPT-6 SolClosed 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 TerravsGPT-6 Solcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins42 – 160 · 227 shared, 25 tied

Comparing matching setups only (9 benchmarks excluded) · suites counted once
← GPT-5.6 TerraGPT-6 Sol →

GPT-5.6 TerravsGPT-6 Solwins by categoryData Nov 15, 2026Data: Vals AI, Mercor, Arena and 50 othersVillage LLM Atlasatlas.village.dev

Margins

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

Vending-Bench 2Games734314428+42.9
SWE-bench Verified Extended (Mercor held-out)Code78.739.3−39.4
BrowseComp Extended (Mercor held-out)Search25.363.2+37.9
Arena Code WebDev — FullstackCode15651703+36.2
Olam Social Poker EloGames15111542+33.7
CadQueryEval (Corrected Geometry)Code68.0100.0+32.0
YC-BenchGames6361441441793+28.1
Terminal-Bench 4.0Code21.549.4+27.9
Gray Swan IPI Attack Success (k=15)Safety37.310.1+27.2
Arena Code WebDev — Reference-Based DesignCode15161720+26.9
LMArena Agent Arena Net ImprovementAgents0.68.8+26.9
GameBench 2Games53.579.4+26.0
Arena Image-to-WebDevCode15391655+25.4
Arena Code WebDev — SimulationsCode15261741+24.3
Arena Code WebDev — Content Creation ToolsCode15031649+23.1
Arena Code WebDev — GamingCode15481751+22.8
ChessBench EloGames9251303+22.5
Arena Code WebDev — HTMLCode15231681+22.1
Arena Code WebDev — ReactCode15111676+22.0
ZeroBench (Main questions, pass@5)Multimodal19.041.0+22.0
DeepsecBench — Revalidated V1Code19.140.9+21.8
DeepsecBench — Revalidated V1 (233 issues)Code19.140.7+21.7
Terminal-Bench Science 0.1 (Vals)Science10.031.4+21.4
Arena Code WebDev — FrontendCode15211687+21.3
DeepsecBench-AA (Artificial Analysis)Cybersecurity25.146.3+21.3
Mystery Game PuzzlesGames35.056.0+21.0
SRE Bench — Fully SolvedCode9.530.5+21.0
LisanBench Difficulty-WeightedGames11973118+20.6
FACTS Benchmark SuiteGeneral QA52.072.5+20.5
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal20.00.0−20.0
ChartographyMultimodal34.053.6+19.6
FrontierMath Tier 4 (v2) - PrivateMath70.790.0+19.3
ChessBench CoherenceGames69.788.2+18.5
AutomationBench 1.0.6 (Zapier)Agents14.833.2+18.4
Arena Code WebDev — Brand & MarketingCode15301657+18.4
Arena Code WebDev — Consumer ProductCode15051649+18.2
GeneBench-Pro Public Release SubsetScience8.026.0+18.0
Arena Code WebDev — Data & AnalyticsCode14961616+18.0
ALE-BenchCode21542637+17.1
MedXpertQA MM Extended (Mercor held-out)Health & Medicine51.068.0+17.0
LMArena Agent Arena Net Improvement (Steering Burden)Agents-1.86.9+16.5
Humanity's Last Exam DiamondGeneral QA28.845.2+16.4
Terminal-Bench-Science 0.1Science13.830.0+16.2
Medical Long Context Reasoning (MLCR-AA)Health & Medicine31.716.1−15.6
SWE Atlas Codebase QnA Extended (Mercor held-out)Code16.031.3+15.3
LisanBench Path LengthGames33038222+15.3
HardGeoBench Avg DistanceMultimodal19911497+15.2
ZeroBench Main Mean pass@1 (five samples)Multimodal14.228.8+14.6
BuseyBench SVGMultimodal6.17.5+14.5
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting67.853.6−14.2
RuneBenchGames5.56.4+14.2
NoCoT-Bench NCRI 15.2General QA89.0115+14.1
APEX-SWECode31.145.0+13.9
Creative Writing v3 EloChat & Writing18562126+13.6
AA-Omniscience IndexIndexes0.127.1+13.5
Vibe Code Bench v1.1Code74.687.8+13.2
SWE-bench MultilingualCode94.481.7−12.7
Legal Research BenchLegal41.328.8−12.5
HealthBench Hard (updated implementation)Health & Medicine34.322.1−12.2
Tax Agent BenchFinance & Accounting65.253.0−12.2
Artificial Analysis Intelligence Index v5.0.1Indexes32.344.2+11.8
Extended NYT Connections 940 (Quadratic v1)Games78.490.1+11.7
FrontierCode 2.0 (Artificial Analysis)Code23.935.6+11.7
HealthBench (updated implementation)Health & Medicine58.747.1−11.6
GDPval Extended (Mercor held-out rubric score)Professional Work47.258.7+11.5
Finance Agent v2 - Disclosure AnalysisFinance & Accounting66.955.6−11.3
CritPt v2 (agentic, avg@5)Science69.780.9+11.1
Roboflow Vision Evals (October 2026, five tasks)Multimodal66.076.9+10.9
FrontierPhysics v0.1Science3.614.3+10.7
SWE-Marathon v1.1Code32.543.1+10.6
NoCoT-Bench NCKI (kspine v3)General QA104119+10.4
LLM Chess EloGames12931545+10.4
Code MigrationCode47.857.2+9.4
Design Arena Overall Code CategoriesCode12381293+9.1
Chess Puzzles (task version 2.0)Games47.056.0+9.0
ProofBench v1.1Math74.083.0+9.0
Terminal-Bench 4.0 (Artificial Analysis)Code35.443.9+8.6
Humanity's Sixth SenseMultimodal22.731.2+8.5
Roboflow Vision Evals (September 2026 protocol)Multimodal73.882.3+8.5
DistressBenchSafety76.468.0−8.5
Financial Coaching BenchFinance & Accounting71.062.6−8.4
Wiz Cyber Model Arena v3.9Cybersecurity63.071.2+8.2
MentalHealthBenchSafety46.254.2+8.0
SUPERChemScience64.972.8+7.9
AA-Briefcase v1.1 EloAgents13331476+7.8
ACCoRDGeneral QA74.782.4+7.7
DAYJOB: FinanceFinance & Accounting1.89.3+7.5
VCT v2 (SecureBio)Science41.949.3+7.4
Conceptual Reasoning IndexIndexes65.771.0+7.4
ZeroBench Main Questions pass^5Multimodal10.017.0+7.0
SM BenchSafety59.066.0+7.0
LiveBench 2.0Agents75.081.9+6.9
VCT (SecureBio)Science47.254.0+6.8
Finance Agent v2 - All PassFinance & Accounting42.836.0−6.8
GBENCH One-shot Coding GScoreGames63.370.1+6.8
Finance Agent v2 - AdjustmentsFinance & Accounting49.743.2−6.5
DeepSWE v1.1Code60.266.6+6.4
Finance Agent v2 - ComparablesFinance & Accounting47.040.8−6.2
Blueprint-Bench 2Multimodal0.30.4+6.1
LAB-Bench CloningScenarios (SecureBio)Science92.598.5+6.0
Terminal-Bench 2.1 Extended (Mercor held-out)Code36.442.4+6.0
LMArena Agent Arena Praise vs ComplaintAgents-1.910.0+5.9
APEX-1 (Mercor, mean score)Professional Work69.563.6−5.9
ChessBench AccuracyGames71.977.4+5.9
CyberBench v1.1 OverallCybersecurity72.178.0+5.9
LMArena Agent Arena Confirmed SuccessAgents-7.24.5+5.9
Public Benefits Bench v1.1Professional Work62.456.6−5.8
ARC-AGI-2General QA83.989.6+5.7
Terminal-Bench 2.1Code77.583.1+5.6
MedXpertQA MM (200-question set)Health & Medicine75.781.3+5.6
Artificial Analysis Intelligence Index v4.3.2Indexes42.147.6+5.5
HieroglyphBenchMultimodal20.526.0+5.5
RCM-Bench-1Health & Medicine87.081.5−5.4
LMArena Agent Arena Steering BurdenAgents-0.510.1+5.3
Excel Modeling BenchmarkFinance & Accounting66.271.5+5.3
LMArena Agent Arena SteerabilityAgents4.114.5+5.2
LMArena Agent Arena Bash ReliabilityAgents0.710.9+5.1
Ruby LLM Benchmarks — Program FixerCode74.079.0+5.1
IOI (Vals v2)Code87.682.6−5.0
Finance Agent v2 - Earnings AnalysisFinance & Accounting77.372.4−4.9
Bio Capabilities Index (SecureBio)Indexes155161+4.9
FrontierCode 1.1 Extended - ScoreCode55.860.7+4.8
Longform Creative WritingChat & Writing78.082.8+4.8
AA-LCR v1.1Search79.083.7+4.7
Agents' Last Exam (ALE)Agents27.632.2+4.6
FrontierCode 1.1 Extended - Pass RateCode61.866.3+4.6
MMMU-Pro Extended (Mercor held-out)Multimodal47.451.9+4.5
Vals Index v2.1Indexes53.157.5+4.5
Humanity's Last Exam (Text-Only)General QA41.946.3+4.4
DeepResearch Bench II (Mercor 130-task set)Search47.242.9−4.3
MBCT (SecureBio)Science56.961.1+4.2
AA-LCR Extended (Mercor held-out)Search74.470.2−4.2
τ³-Banking (Mercor run, bm25_grep retrieval)Agents38.843.0+4.2
Furniture Assembly BenchmarkMultimodal54.258.3+4.2
LMCAGeneral QA55.059.1+4.1
LAB-Bench ProtocolQA (SecureBio)Science77.081.1+4.1
DTBenchGeneral QA93.397.3+4.0
FrontierMath Tiers 1-3 (v2) - PrivateMath86.089.8+3.9
ARC-AGI-3 (Semi-Private)General QA0.84.6+3.8
HealthBench (length-adjusted)Health & Medicine57.053.2−3.8
Harvey HLAB — Criteria Pass RateLegal79.883.5+3.8
Finance Agent v2 - Financial ModelingFinance & Accounting23.019.3−3.7
MedCodeHealth & Medicine43.447.1+3.7
Finance Agent v2 - PrecedentsFinance & Accounting24.828.5+3.7
BrowseComp (Mercor 130-question subset)Search85.889.4+3.6
GPQA-Bio (SecureBio)Science85.388.9+3.6
Artificial Analysis Progress Index v1Indexes324334+3.6
Quant-Arena (iterative total score)Finance & Accounting47.651.1+3.6
SciCode (main problems)Science15.812.7−3.1
APEX-Agents 1.1Professional Work71.868.7−3.1
LMArena Agent Arena Bash RecoveryAgents0.97.0+3.1
MMMU ProMultimodal79.582.5+3.1
Eval Connections One-Shot with TrapsGames86.089.0+3.0
Vals Index v2Indexes59.662.6+3.0
HealthBench ProfessionalHealth & Medicine62.459.5−2.9
SAB RankingForecasting-0.0-0.1−2.8
AdvancedIFChat & Writing78.180.9+2.8
GDPval (Mercor 213-task rubric score)Professional Work80.783.5+2.8
CharXiv Extended (Mercor held-out)Multimodal81.183.8+2.7
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting72.770.0−2.7
Finance Agent v2 - Market AnalysisFinance & Accounting60.658.0−2.7
SciCode 1.0.1 (Artificial Analysis)Code55.057.6+2.7
HealthBench Hard (length-adjusted)Health & Medicine32.730.1−2.6
Bug Hunt BenchCode30.527.9−2.5
Epoch Capabilities Index (ECI)Indexes161164+2.5
SWE-bench VerifiedCode77.880.3+2.5
GDPval-AA v2.1Professional Work14611514+2.3
VulcanBench Frontier v4Code89.186.8−2.2
SAB ResearchFinance & Accounting93.495.6+2.2
SAGEMath47.044.8−2.2
AutomationBench-AA 1.0.6Agents59.661.7+2.0
DAYJOB: HealthcareHealth & Medicine1.63.6+2.0
LMArena Text Arena EloChat & Writing14681457−1.9
CVDP (517-problem set)Code50.952.7+1.8
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal2.34.1+1.8
ITBench-AAAgents51.049.4−1.6
RiemannBenchMath66.464.8−1.6
ProgramBenchCode0.52.0+1.5
ObviousBenchGeneral QA97.999.3+1.4
HardGeoBench <=100miMultimodal36.337.5+1.3
HardGeoBench <=10miMultimodal25.026.3+1.3
HardGeoBench Country AccuracyMultimodal45.046.3+1.3
SWE Atlas - Codebase QnACode45.244.1−1.1
ARC-AGI-1 Semi-PrivateGeneral QA96.595.5−1.0
CharXiv (descriptive and reasoning combined)Multimodal94.195.0+0.9
CritPt (avg@5)Science30.030.9+0.9
MedScribeHealth & Medicine82.982.0−0.8
MMLU ProGeneral QA88.188.9+0.8
Creative Writing v3 Rubric ScoreChat & Writing15.916.1+0.8
LAB-Bench LitQA2 (SecureBio)Science51.151.8+0.7
WCB (SecureBio)Science59.860.5+0.7
GDP.pdf (Artificial Analysis, All-pass)Professional Work24.625.2+0.6
HealthBench Consensus (updated implementation)Health & Medicine95.295.8+0.6
PubMedQA (SecureBio)Health & Medicine80.480.9+0.5
SpeciEval (Pooled English Runs)Safety98.198.5+0.4
SpeciEval (Refusal-Aware Retries, Historical)Safety98.198.5+0.4
HPCT (SecureBio)Science68.267.8−0.4
WMDP-Bio (SecureBio)Science89.790.0+0.3
OTIS Mock AIME 2024-2025Math99.7100.0+0.3
APEX-Accounting 1.1Finance & Accounting59.859.7−0.1
MMLU-Bio (SecureBio)Science90.790.8+0.1
LMArena Agent Arena Tool HallucinationAgents1.01.00.0
AIME 2025Math100.0100.00.0
BullshitBench v2 LeaderboardSafety54.054.00.0
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 - 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 - Energy Natural ResourcesLegal0.00.00.0
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

GPT-5.6 TerravsGPT-6 Solnormalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, Arena and 50 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…