Atlas

Compare Models

Claude Opus 5 vs GPT-5.5

Claude Opus 5 leads by 4.1 capability points (≈95% interval +1.9 to +6.2): a clear lead over GPT-5.5 · P(higher) ≈ 100%

Wins 194 of 270 shared benchmarks (46 lost, 30 tied)

Capability vs. relative cost
Claude Opus 5GPT-5.5Closed 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.

Claude Opus 5vsGPT-5.5capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins194 – 46 · 270 shared, 30 tied

Comparing matching setups only (85 benchmarks excluded) · suites counted once
← Claude Opus 5GPT-5.5 →

Claude Opus 5vsGPT-5.5wins by categoryData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 74 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-5.5 minus Claude Opus 5.

KernelBench Hard Normalized (H100)Code82.80.0−82.8
CyberBench PoCCybersecurity0.079.7+79.7
KernelBench Mega Normalized (H100)Code100.023.1−76.9
HPCT Refusal (CAIS)Safety100.029.0−71.0
Machiavelli Deception (CAIS)Safety96.581.7+44.7
SWE-bench Verified Extended (Mercor held-out)Code82.039.3−42.7
Medical Long Context Reasoning (MLCR-AA)Health & Medicine59.417.8−41.7
CyberBench OverallCybersecurity40.780.5+39.8
ChessBench CoherenceGames63.8100.0+36.2
TasteVal (compute multiplier)Science1.30.5−35.8
Terminal-Bench 4.0 (Artificial Analysis)Code46.514.6−31.8
ChessBench EloGames8261349+31.1
ARC-AGI-3 (Semi-Private)General QA30.20.4−29.7
AA-Briefcase EloAgents16251136−28.4
ProofBenchMath78.050.0−28.0
LLM Debate BenchmarkChat & Writing17241528−28.0
JobBench MainProfessional Work65.738.3−27.4
APEX-SWECode63.736.6−27.1
VoxelBench Text RatingGames23001925−26.7
BioTIER RefusalSafety100.073.4−26.6
BullshitBench v2 LeaderboardSafety73.047.0−26.0
ScBench-LongScience41.315.9−25.4
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine40.014.7−25.3
BrowseComp Extended (Mercor held-out)Search44.019.0−25.0
GameBench 2Games84.859.9−24.9
EBR-bench v2Games45.721.0−24.8
YC-BenchGames19115921206045−24.6
OSWorld 2.0 500-step Partial ScoreAgents70.647.5−23.1
Drone-BenchAgents86.963.9−23.0
SWE Atlas Codebase QnA Extended (Mercor held-out)Code40.017.3−22.7
Vending-Bench 2Games111827524−22.1
ExploitBench - release-reported exploit progressCybersecurity70.047.9−22.1
Document ArenaProfessional Work15091484−21.8
BrokenArXiv 06/2026Math90.769.4−21.3
SWE-Bench Pro (Public Dataset)Code79.258.6−20.6
SM BenchSafety85.865.4−20.4
VariantBench V1 Clean RerunsScience50.630.5−20.0
Chess Puzzles (task version 2.0)Games43.063.0+20.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal20.00.0−20.0
VariantBenchScience49.429.7−19.7
ClockBench AccuracyMultimodal65.546.1−19.4
LisanBench Difficulty-WeightedGames50683316−18.8
Vibe Code Bench v1.1Code88.469.8−18.6
ALE-BenchCode26222108−18.2
TxBench Antibody Discovery V0Science51.433.6−17.8
BankerToolBench v1.1Finance & Accounting32.615.0−17.6
LMArena Agent Arena Net ImprovementAgents9.54.2−17.5
EnterpriseBench: CoreCraftProfessional Work68.751.3−17.4
LisanBench Path LengthGames154289845−17.3
HiL-Bench (Human-in-Loop Benchmark)Agents57.039.7−17.3
Humanity's Last Exam DiamondGeneral QA44.827.9−16.9
GDPval-AA v2 (normalized)Professional Work66.449.5−16.8
Furniture Assembly BenchmarkMultimodal60.844.2−16.7
Tax Agent BenchFinance & Accounting76.760.5−16.2
Public Benefits Bench v1.1Professional Work76.960.9−16.0
Antidote: Everyday EditionSafety10551013−15.0
Legal Research BenchLegal55.340.4−14.9
GDPval-AA v2.1Professional Work16951357−14.9
Bio Capabilities Index (SecureBio)Indexes178158−14.8
Creative Writing v3 EloChat & Writing21351845−14.7
LLM Creative Story WritingChat & Writing3.72.4−14.7
MedCodeHealth & Medicine63.649.1−14.5
SWE-bench VerifiedCode97.082.6−14.4
BioSecBench-Function V1 Clean RerunsScience42.528.3−14.2
GeneBench-Pro Public Release SubsetScience22.08.0−14.0
Harvey LAB (Mercor run, 1,749 public tasks)Legal17.84.1−13.7
ScBench-Long V0 VerifiedScience36.422.7−13.6
AURORA Multilingual Terminal-BenchCode73.760.3−13.4
VCT-Refusal (CAIS)Safety29.416.0−13.4
LiveBench 2.0Agents82.168.7−13.4
MineBench Conservative RatingGames19951772−13.2
RiemannBenchMath68.055.2−12.8
GDPval-AA v2Professional Work16291372−12.7
Humanity's Last ExamGeneral QA56.443.8−12.6
Artificial Analysis Intelligence Index v4.3.2Indexes50.838.4−12.4
CyberGymCybersecurity94.281.8−12.4
Code MigrationCode57.545.2−12.3
ComplexConstraintsChat & Writing37.349.5+12.2
AutomationBench (Overall)Agents25.112.9−12.2
Artificial Analysis Intelligence Index v4.3Indexes50.738.6−12.1
TxBench Antibody Discovery V1 Clean RerunsScience41.929.9−11.9
HPCT (SecureBio)Science78.666.8−11.8
CursorBench 3.2Code70.058.4−11.6
SimpleBenchGeneral QA80.669.0−11.6
DeepsecBenchCode32.621.2−11.4
DeepsecBench — Revalidated V1Code32.421.1−11.3
DeepsecBench — Revalidated V1 (233 issues)Code32.321.0−11.3
HANDBOOK.mdProfessional Work32.321.5−10.8
DeepResearch Bench II (Mercor 130-task set)Search56.145.4−10.7
Design Arena Overall Code CategoriesCode13211257−10.6
BabyVision with toolsMultimodal94.183.6−10.5
MCP Atlas April 2026 (all 1,000 tasks)Agents85.875.3−10.5
Time Horizon Index: KSPAgents18.88.3−10.5
Artificial Analysis Agentic IndexIndexes55.344.9−10.4
AutomationBench 1.0.6 (Zapier)Agents26.916.6−10.4
SpeciEval (Pooled English Runs)Safety87.998.2+10.3
SpeciEval (Refusal-Aware Retries, Historical)Safety87.998.2+10.3
LMCAGeneral QA64.554.3−10.2
LLM Chess EloGames12851532+10.1
Conceptual Reasoning IndexIndexes74.767.3−10.1
MineBench — Bradley-Terry RatingGames20081846−9.8
EpiBenchScience35.245.0+9.8
MMMU-Pro Extended (Mercor held-out)Multimodal34.343.8+9.5
ATLAS-Education (StudentBench) Lesson PlanningProfessional Work1.10.9−9.4
AutomationBench-AA 1.0.6Agents56.647.3−9.3
GlobeBenchGeneral QA15.925.2+9.3
Agents' Last Exam (ALE)Agents32.223.0−9.2
WCB (SecureBio)Science68.859.7−9.1
Excel Modeling BenchmarkFinance & Accounting73.664.5−9.0
Judgemark v4Chat & Writing78.887.8+9.0
TxBench Oligo Discovery V0Science42.233.6−8.5
DeepSWE v1.1Code72.864.4−8.4
GDPval Extended (Mercor held-out rubric score)Professional Work62.954.5−8.4
ChessBench AccuracyGames71.078.8+8.4
AA-Omniscience IndexIndexes37.120.5−8.3
HieroglyphBenchMultimodal14.622.8+8.2
NoCoT-Bench NCRI 15.2General QA115100−8.2
Longform Creative WritingChat & Writing86.378.2−8.1
TxBench Oligo Discovery V1 Clean RerunsScience40.031.9−8.1
SUPERChemScience74.766.8−7.9
Harvey HLAB — Criteria Pass RateLegal87.780.0−7.7
BioSecBench-RefusalSafety29.437.1+7.7
TxBench Preclinical Pharmacology V1 Clean RerunsScience63.055.3−7.7
ExploitGymCybersecurity22.015.1−6.9
WeirdML (v2)Science91.884.9−6.9
DeepSearchQA (F1 score)Search94.687.8−6.8
Finance Agent v2Finance & Accounting58.651.8−6.8
APEX-Agents 1.1Professional Work77.070.3−6.7
EnigmaEvalGeneral QA43.937.2−6.7
BioSecBench-Function V0Science43.036.5−6.5
BrowseCompSearch90.884.4−6.4
APEX-AccountingFinance & Accounting54.548.2−6.3
Tampered MLScience77.971.8−6.1
NoCoT-Bench NCKI (kspine v3)General QA119111−6.0
CorpFin v2 - Max Fitting ContextFinance & Accounting76.070.0−5.9
GBENCH One-shot Coding GScoreGames68.162.2−5.9
Blueprint-Bench 2Multimodal0.30.4+5.8
𝜏³-BankingAgents44.739.0−5.8
MRCR v2 (8-needle, Context Arena) - 128K AUCGeneral QA97.591.7−5.8
ScBench-Long V1 Clean RerunsScience27.321.6−5.7
MRCR v2 (8-needle, Context Arena) - 128K pointwiseGeneral QA91.385.8−5.5
ARC-AGI-2General QA90.485.0−5.4
CorpFin v2 - Shared Max ContextFinance & Accounting74.969.6−5.4
CorpFin v2 (Vals Index component)Finance & Accounting74.969.6−5.4
CorpFin v2 (Vals Multimodal Index component)Finance & Accounting74.969.6−5.4
IntPhys 2Multimodal65.159.9−5.2
Senior SWE-BenchCode34.729.5−5.2
Ramp SWE-Bench (78 published tasks)Code88.583.3−5.1
A-Fantasia (Valid Attempts with Retries)Games29.734.7−5.0
SciCode (main problems)Science17.312.3−5.0
Roboflow Vision Evals (October 2026, five tasks)Multimodal70.966.0−4.9
Artificial Analysis Progress Index v1Indexes334320−4.8
GPQA-Bio (SecureBio)Science82.787.4+4.7
OSWorld-VerifiedAgents83.478.7−4.7
TextQuests Game ProgressGames53.348.9−4.4
MindCube TinyMultimodal84.379.9−4.4
BioSecBench-Surveillance V1 Clean RerunsScience51.547.1−4.4
SpatialBench V1 Clean RerunsScience68.464.1−4.3
TextQuests Average Cumulative HarmSafety22.621.4+4.2
SpatialBench-Long V1 Clean RerunsScience28.032.1+4.2
MedScribeHealth & Medicine91.086.9−4.1
CadQueryEval (Corrected Geometry)Code80.076.0−4.0
TaxCalcBench TY25Finance & Accounting20.024.0+4.0
ExploitGym - release-reported percentage, cohort unspecifiedCybersecurity22.025.9+3.9
AA-AnalystAgent (pass^5)Professional Work53.850.0−3.7
Terminal-Bench 2.1Code88.084.3−3.7
LiveCodeBench (Vals AI public dataset)Code89.085.3−3.7
CritPt (avg@5)Science29.125.4−3.7
Epoch Capabilities Index (ECI)Indexes164160−3.6
Roboflow Vision Evals (September 2026 protocol)Multimodal78.374.8−3.5
PG-LLM-50Science0.40.3−3.5
AA-LCR Extended (Mercor held-out)Search73.970.5−3.4
ChessBench — Historical Output RepairGames32.729.3−3.3
AA-LCRSearch70.073.3+3.3
InteractiveBench (task pass)Agents33.330.0−3.3
ACCoRDGeneral QA83.880.5−3.3
Hemingway-benchChat & Writing10471035−3.3
MASKSafety93.490.1−3.3
ProgramBench (Mercor 100-task subset)Code4.31.0−3.3
MedXpertQA MM Extended (Mercor held-out)Health & Medicine64.160.8−3.3
PCT Political Inconsistency (CAIS)Safety36.740.0−3.3
MortgageTax (Vals all 1,258 documents)Finance & Accounting72.168.8−3.3
Olam Social Poker EloGames15181521+3.3
SimpleQA VerifiedGeneral QA59.963.1+3.2
LAB-Bench ProtocolQA (SecureBio)Science81.177.9−3.2
EpiBench V1 Clean RerunsScience25.228.3+3.1
MedXpertQA MM (200-question set)Health & Medicine79.876.7−3.1
MMLU ProGeneral QA92.088.9−3.1
MBCT (SecureBio)Science60.257.1−3.1
Artificial Analysis Coding IndexIndexes78.074.9−3.1
GPQA DiamondGeneral QA92.495.5+3.1
CorpFin v2 - Exact PagesFinance & Accounting68.665.6−3.0
BuseyBench SVGMultimodal6.36.0−3.0
Mystery Game PuzzlesGames59.056.0−3.0
CharXiv Extended (Mercor held-out)Multimodal79.882.7+2.9
MMMU ProMultimodal84.081.2−2.8
LAB-Bench CloningScenarios (SecureBio)Science100.097.3−2.7
ARC-AGI-1 Semi-PrivateGeneral QA97.595.0−2.5
AA-LCR v1.1Search82.084.3+2.3
Ruby LLM Benchmarks — Program FixerCode79.877.6−2.2
DAXBench v2Code86.288.3+2.1
SAGEMath49.451.5+2.1
LMArena Agent Arena Bash RecoveryAgents13.69.5−2.0
GDP.pdfProfessional Work24.026.0+2.0
LMArena Agent Arena SteerabilityAgents10.76.9−1.9
Extended NYT Connections 940 (Quadratic v1)Games94.396.2+1.9
DTBenchGeneral QA97.996.0−1.9
Creative Writing v3 Rubric ScoreChat & Writing16.315.9−1.9
Agent Red Teaming k=100 (CAIS)Safety43.341.5+1.8
SpatialViz-BenchMultimodal75.874.2−1.6
LAB-Bench LitQA2 (SecureBio)Science51.449.8−1.6
Bench to the Future 3 — Pooled ScoreForecasting0.10.1−1.6
SpatialBench-Long V0 VerifiedScience31.833.3+1.5
LMArena Text Arena EloChat & Writing14901481−1.5
SpatialBench V0 VerifiedScience66.765.2−1.5
CRM-Agent-BenchAgents82.781.3−1.4
ForecastBench Preliminary Dataset Brier IndexForecasting62.063.3+1.3
ScBench V1 Clean RerunsScience58.357.1−1.2
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
GDPval (Mercor 213-task rubric score)Professional Work85.684.6−1.0
WordleBenchGames96.097.0+1.0
VCT (SecureBio)Science51.452.3+0.9
EnterpriseOps-Gym-AAAgents47.546.6−0.8
AdvancedIFChat & Writing80.581.3+0.8
FrontierMath Tier 4 (v2) - PrivateMath73.272.5−0.7
Toloka Arena (v1)Agents52.952.3−0.6
matplotgymCode56.555.9−0.6
Vision Arena OverallMultimodal12851283−0.5
LegalBenchLegal87.086.5−0.5
ATLAS-Education (StudentBench) Learning GainProfessional Work12.613.5+0.5
ForecastBench Tournament Overall Brier IndexForecasting66.667.0+0.4
FrontierMath Tiers 1-3 (v2) - PrivateMath85.685.3−0.4
ERQAMultimodal70.870.5−0.3
SciCode (Artificial Analysis avg@3)Science55.755.9+0.2
APEX-1 (Mercor, mean score)Professional Work68.668.4−0.2
TaxEval v2Finance & Accounting75.175.0−0.2
RedlineBenchLegal44.044.1+0.1
ActiveVisionMultimodal10.610.60.0
AIME 2025Math100.0100.00.0
BabaIsBench Stage 0: The IntroGames100.0100.00.0
CharXiv (descriptive and reasoning combined)Multimodal94.694.60.0
CyberBench PatchCybersecurity81.481.40.0
FrontierMath ErdősMath0.00.00.0
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 - Emerging Companies Venture CapitalLegal0.00.00.0
Harvey's Legal Agent Benchmark - Employment LaborLegal0.00.00.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal20.020.00.0
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 - 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
ObviousBenchGeneral QA100.0100.00.0
Terminal-Bench 2.1 Extended (Mercor held-out)Code38.038.00.0
ZeroBench Main Mean pass@1 (five samples)Multimodal17.217.20.0
ZeroBench Main Questions pass^5Multimodal11.011.00.0

Claude Opus 5vsGPT-5.5normalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 74 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…