Atlas

Compare Models

Claude Opus 5 vs Claude Sonnet 5.5

Claude Sonnet 5.5 leads by 2.5 capability points (≈95% interval +0.0 to +5.0): a clear lead over Claude Opus 5 · P(higher) ≈ 98%

Wins 125 of 205 shared benchmarks (59 lost, 21 tied)

Capability vs. relative cost
Claude Opus 5Claude Sonnet 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 5vsClaude Sonnet 5.5capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins59 – 125 · 205 shared, 21 tied

Comparing matching setups only (12 benchmarks excluded) · suites counted once
← Claude Opus 5Claude Sonnet 5.5 →

Claude Opus 5vsClaude Sonnet 5.5wins by categoryData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 60 othersVillage LLM Atlasatlas.village.dev

Margins

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

KernelBench CUDA Normalized (RTX PRO 6000)Code45.587.0+41.6
ActiveVisionMultimodal10.641.2+30.6
ARC-AGI-3 (Semi-Private)General QA30.21.2−29.0
Terminal-Bench-Science 0.1Science28.653.3+24.8
SWE-bench Verified Extended (Mercor held-out)Code82.057.3−24.7
Bug Hunt BenchCode25.748.9+23.2
LisanBench Difficulty-WeightedGames50682983−22.3
Concept Synth INDUCTION Challenge100 Round 1General QA24.044.0+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 - Energy Natural ResourcesLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal0.020.0+20.0
Terminal-Bench Science 0.1 (Vals)Science27.147.1+20.0
Harvey HLAB — Criteria Pass RateLegal87.770.0−17.7
Arena Image-to-WebDevCode16641744+17.5
Concept Synth INDUCTION Challenge64 Round 1General QA32.850.0+17.2
GeneBench-Pro Public Release SubsetScience22.039.0+17.0
VIALSMultimodal24.441.0+16.6
HardGeoBench Avg DistanceMultimodal1218683+16.5
LMArena Agent Arena Net Improvement (Steering Burden)Agents7.216.0+16.5
Chess Puzzles (task version 2.0)Games43.059.0+16.0
FrontierCode 2.0 - Pass RateCode30.446.4+15.9
Medical Long Context Reasoning (MLCR-AA)Health & Medicine59.475.0+15.6
AutomationBench-AA 1.0.6Agents56.671.8+15.2
FrontierCode 2.0 - ScoreCode27.242.2+15.0
APEX-1 (Mercor, mean score)Professional Work68.653.9−14.7
Terminal-Bench 4.0 (Artificial Analysis)Code49.063.6+14.6
Furniture Assembly BenchmarkMultimodal60.875.0+14.2
Extended NYT Connections 940 (Quadratic v1)Games94.380.5−13.8
BullshitBench v2 LeaderboardSafety73.060.0−13.0
DAYJOB: FinanceFinance & Accounting11.324.0+12.7
Code MigrationCode57.569.8+12.4
MysteryMechanismScience37.449.1+11.7
NoCoT-Bench NCRI 15.2General QA11595.1−11.0
LMArena Code Arena WebDevCode16911773+10.7
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine40.029.3−10.7
MedCodeHealth & Medicine63.652.9−10.7
Artificial Analysis Intelligence Index v5.0.1Indexes44.254.5+10.4
Humanity's Last ExamGeneral QA56.446.1−10.3
τ³-Banking (Mercor run, bm25_grep retrieval)Agents44.054.0+10.0
FrontierSWE V2Code52.061.9+9.9
Public Benefits Bench v1.1Professional Work76.967.2−9.7
TxBench Antibody Discovery V1 (100 evaluations)Science48.057.3+9.3
LMArena Agent Arena Praise vs ComplaintAgents12.230.5+9.2
SPU (Science Prediction Universe)Science0.20.3+8.9
Blueprint-Bench 2Multimodal0.30.4+8.9
CursorBench 4.0Code46.655.5+8.9
AA-Briefcase v1.1 EloAgents16541816+8.9
BioSecBench-Function V1 Clean RerunsScience44.853.5+8.7
Olam Social Poker EloGames15181526+8.7
Artificial Analysis Coding Agent Index v1.5Indexes59.768.4+8.6
Humanity's Last Exam DiamondGeneral QA44.853.3+8.5
MedXpertQA MM Extended (Mercor held-out)Health & Medicine64.155.6−8.5
NoCoT-Bench NCKI (kspine v3)General QA119107−8.5
GDPval Extended (Mercor held-out rubric score)Professional Work62.970.9+8.0
Terminal-Bench 4.0Code53.961.8+7.9
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting49.241.5−7.7
Terminal-Bench 2.1 Extended (Mercor held-out)Code38.045.5+7.5
GDP.pdfProfessional Work24.031.4+7.4
FrontierMath Tier 4 (v2) - PrivateMath73.280.5+7.3
FrontierCode 1.1 Extended - Pass RateCode62.669.9+7.3
LAB-Bench ProtocolQA (SecureBio)Science81.188.3+7.2
DAYJOB: HealthcareHealth & Medicine8.415.2+6.8
APEX-Agents 1.1Professional Work77.083.7+6.7
Roboflow Vision Evals (October 2026, five tasks)Multimodal70.977.6+6.7
GameBench 2Games84.891.4+6.6
ARC-AGI-2General QA90.483.9−6.5
ALE-BenchCode26222448−6.2
OSWorld 2.1 full - reported partial reward, step limit unspecifiedAgents74.080.1+6.1
Ruby LLM Benchmarks — Program FixerCode79.873.8−6.0
MetagenomicsBench V0Science54.060.0+6.0
MetagenomicsBench V1Science54.060.0+6.0
Mystery Game PuzzlesGames59.065.0+6.0
YC-BenchGames19115921741315−5.9
Artificial Analysis Progress Index v1Indexes334351+5.9
Vision Arena OverallMultimodal12851263−5.8
SpatialBench-Long V1 Clean RerunsScience30.936.5+5.6
DeepSWE v1.1Code65.070.5+5.5
Roboflow Vision Evals (September 2026 protocol)Multimodal78.383.8+5.5
Artificial Analysis Intelligence Index v4.3.2Indexes50.856.0+5.2
DeepsecBench — Revalidated V1 (233 issues)Code32.337.5+5.2
ChessBench AccuracyGames71.075.8+5.1
LisanBench Path LengthGames1542813787−5.1
LMArena Agent Arena Steering BurdenAgents4.414.5+5.1
EpiBench V1 Clean RerunsScience25.230.2+5.0
HardGeoBench Country AccuracyMultimodal53.858.8+5.0
GPQA-Bio (SecureBio)Science82.787.6+4.9
GDPval-AA v2.1Professional Work17291838+4.8
SWE Atlas Codebase QnA Extended (Mercor held-out)Code40.044.7+4.7
SimpleBenchGeneral QA80.675.9−4.7
LMArena Agent Arena Confirmed SuccessAgents10.219.5+4.7
Design Arena Overall Code CategoriesCode13211349+4.6
SciCode 1.0.1 (Artificial Analysis)Code56.461.0+4.6
ExploitGymCybersecurity22.026.6+4.6
CharXiv Extended (Mercor held-out)Multimodal79.884.2+4.4
DeepResearch Bench II (Mercor 130-task set)Search56.151.7−4.4
LAB-Bench LitQA2 (SecureBio)Science51.455.8+4.4
DistressBenchSafety70.274.6+4.4
SpatialBench V1 Clean RerunsScience68.472.8+4.3
CritPt v2 (agentic, avg@5)Science71.475.7+4.3
Vibe Code Bench v1.1Code88.492.4+4.0
LMCAGeneral QA64.568.3+3.8
HardGeoBench <=100miMultimodal46.350.0+3.8
HardGeoBench <=10miMultimodal32.536.3+3.8
AA-AnalystAgent (pass^5)Professional Work53.857.5+3.7
LLM Chess EloGames12851195−3.7
RedlineBenchLegal44.040.3−3.7
TxBench Oligo Discovery V1 (113 evaluations)Science48.451.9+3.5
Physical-Engineering's Last Exam (hidden test)Professional Work41.544.9+3.4
MaxxBenchCode63.566.9+3.4
Conceptual Reasoning IndexIndexes74.777.1+3.3
FrontierMath Tiers 1-3 (v2) - PrivateMath85.688.8+3.2
SpeciEval (Pooled English Runs)Safety87.991.0+3.1
SpeciEval (Refusal-Aware Retries, Historical)Safety87.991.0+3.0
ScBench-Long V1 Clean RerunsScience27.330.3+3.0
LMArena Text Arena EloChat & Writing14901472−3.0
LAB-Bench CloningScenarios (SecureBio)Science100.097.0−3.0
AdvancedIFChat & Writing80.583.4+2.9
FrontierMath ErdősMath0.02.9+2.9
RuneBenchGames5.75.5−2.8
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science14.617.3+2.7
APEX-SWECode63.766.4+2.7
PLE (Physical-Engineering's Last Exam)Professional Work42.645.2+2.6
HealthBench ProfessionalHealth & Medicine73.470.8−2.6
SWE-Marathon v1.1Code48.851.3+2.5
Bio Capabilities Index (SecureBio)Indexes178175−2.4
AA-Omniscience IndexIndexes37.132.3−2.4
SAGEMath49.451.8+2.3
AA-LCR v1.1Search82.079.7−2.3
BuseyBench SVGMultimodal6.36.5+2.3
WCB (SecureBio)Science68.866.5−2.3
CritPt (avg@5)Science29.131.4+2.3
Excel Modeling BenchmarkFinance & Accounting73.675.7+2.2
EnterpriseBench: CoreCraftProfessional Work68.770.8+2.1
ScBench V1 Clean RerunsScience58.360.3+2.1
BrowseComp (Mercor 130-question subset)Search84.686.6+2.0
Eval Connections One-Shot with TrapsGames74.076.0+2.0
TaxCalcBench TY25Finance & Accounting20.018.0−2.0
Toloka Arena (v3)Agents62.064.0+2.0
InferenceBenchCode8.99.1+1.9
ChessBench EloGames826857+1.8
ACCoRDGeneral QA83.885.7+1.8
MMMU ProMultimodal84.085.8+1.8
Terminal-Bench 2.1Code89.187.3−1.8
micro1 PharmaBenchHealth & Medicine9.711.4+1.7
VariantBench V1 Clean RerunsScience52.851.1−1.7
Clipping Bench v1.0Games734886+1.6
RCM-Bench-1Health & Medicine92.791.1−1.6
CVDP (517-problem set)Code62.260.6−1.6
CharXiv (descriptive and reasoning combined)Multimodal94.696.1+1.5
Epoch Capabilities Index (ECI)Indexes164166+1.5
LMArena Agent Arena Bash ReliabilityAgents9.812.7+1.5
SWE-TogetherCode63.862.4−1.4
MBCT (SecureBio)Science60.258.8−1.4
HPCT (SecureBio)Science78.677.2−1.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience63.064.3+1.3
IOI (Vals v2)Code84.383.1−1.3
HealthBenchHealth & Medicine67.165.9−1.2
BioSecBench-Structure V1 (58 evaluations)Science20.721.8+1.1
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
VCT (SecureBio)Science51.452.5+1.1
ARC-AGI-1 Semi-PrivateGeneral QA97.596.5−1.0
MMLU ProGeneral QA92.091.0−1.0
ProofBench v1.1Math99.0100.0+1.0
WeirdML v3Science19.019.9+0.9
GDPval (Mercor 213-task rubric score)Professional Work85.686.5+0.9
MedXpertQA MM (200-question set)Health & Medicine79.880.7+0.9
ForecastBench Preliminary Dataset Brier IndexForecasting62.061.1−0.9
ChessBench CoherenceGames63.864.6+0.8
GBENCH One-shot Coding GScoreGames68.167.4−0.8
LMArena Agent Arena Bash RecoveryAgents13.615.0+0.7
MineBench — Bradley-Terry RatingGames20081998−0.6
DTBenchGeneral QA97.997.3−0.5
Harvey LAB-AA v1.0 (criterion pass rate)Legal93.593.1−0.4
LiveBench 2.0Agents82.182.5+0.4
SM BenchSafety85.885.4−0.4
APEX-Accounting 1.1Finance & Accounting64.364.5+0.2
MedScribeHealth & Medicine91.091.1+0.1
Vending-Bench 2Games1118211165−0.1
AA-LCR Extended (Mercor held-out)Search73.973.8−0.1
MMLU-Bio (SecureBio)Science91.191.0−0.1
LMArena Agent Arena Tool HallucinationAgents0.90.90.0
AIME 2025Math100.0100.00.0
DAXBench v2Code86.286.20.0
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 - 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 - White Collar Defense InvestigationsLegal0.00.00.0
ObviousBenchGeneral QA100.0100.00.0

Claude Opus 5vsClaude Sonnet 5.5normalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 60 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…