Atlas

Compare Models

Claude Sonnet 5.5 vs Gemini 4 Argon

Claude Sonnet 5.5 leads by 0.7 capability points (≈95% interval −1.9 to +3.3): statistically tied · P(higher) ≈ 70%

Wins 117 of 234 shared benchmarks (101 lost, 16 tied)

Capability vs. relative cost
Claude Sonnet 5.5Gemini 4 ArgonClosed 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 Sonnet 5.5vsGemini 4 Argoncapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins117 – 101 · 234 shared, 16 tied

Comparing matching setups only (38 benchmarks excluded) · suites counted once
← Claude Sonnet 5.5Gemini 4 Argon →

Claude Sonnet 5.5vsGemini 4 Argonwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio and 60 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Argon minus Claude Sonnet 5.5.

AA-PersonalityBench Transparency indexChat & Writing10021.9−78.8
ARC-AGI-3 (Semi-Private)General QA35.597.3+61.8
AA-PersonalityBench Slop indexChat & Writing27.767.6+40.4
Harvey's Legal Agent Benchmark - Capital MarketsLegal0.040.0+40.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal0.040.0+40.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.040.0+40.0
AA-PersonalityBench Imagination indexChat & Writing84.846.7−38.5
ChessBench EloGames8571485+37.4
Terminal-Bench-Science 0.1Science16.248.6+32.4
NoCoT-Bench NCRI 15.2General QA95.1151+30.8
ChessBenchGames39.370.0+30.7
Terminal-Bench 2.1Code87.357.7−29.6
SimpleQA Verified (Epoch anti-abstention)General QA46.576.0+29.5
SAFE-Teen 106 worst branch: Urgent Safety (no prompt)Safety23.252.6−29.4
AA-PersonalityBench Organization indexChat & Writing84.856.2−28.8
SAFE-Teen 106 worst branch: Relationships (no prompt)Safety37.065.7−28.7
ChessBench CoherenceGames64.692.6+28.0
SAFE-Teen 106 worst branch: Health Advice (no prompt)Safety42.970.8−28.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine41.713.9−27.8
TaxCalcBench TY25 (web search)Finance & Accounting48.074.0+26.0
Terminal-Bench 4.0 (Artificial Analysis)Code29.855.1+25.3
SAFE-Teen 106: Health Advice (no prompt)Safety33.257.4−24.2
Harvey HLAB — Criteria Pass RateLegal70.094.0+24.1
sudo L7Code41.718.9−22.8
HoneyBench v0.2 (hack rate)Safety41.663.6−22.0
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - ImmigrationLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Intellectual PropertyLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - International Trade SanctionsLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal20.040.0+20.0
LisanBench Path LengthGames137877560−19.3
Ramp Accounting BenchFinance & Accounting42.561.1+18.7
SAFE-Teen 106: Urgent Safety (no prompt)Safety14.733.1−18.4
SAFE-Teen 106 worst branch: Urgent Safety (under-18 prompt)Safety11.830.2−18.4
LMArena Agent Arena Bash RecoveryAgents15.0-21.4−18.2
TreasuryQA BenchProfessional Work41.960.0+18.1
Visa Agentic Commerce BenchmarkAgents64.482.2+17.8
MetagenomicsBench V1Science60.042.3−17.7
APEX-SWECode66.449.2−17.2
DAYJOB: AccountingFinance & Accounting35.218.0−17.2
DAYJOB: LawLegal27.09.9−17.1
LLM Chess EloGames11951609+17.0
IOI (Vals v2)Code83.1100.0+16.9
HLAB-Vals-1.1Legal10.427.1+16.7
VibeVideoBench (overall quality, no video = 0)Code3.93.1−16.2
Code Migration v1.1Code86.970.8−16.1
Eval Connections One-Shot with TrapsGames76.092.0+16.0
FrontierCode 2.0 - Pass RateCode46.430.6−15.7
Concept Synth INDUCTION Challenge64 Round 1General QA50.065.6+15.6
Blueprint-Bench 2Multimodal0.40.5+15.1
Code2Video-AA EloCode14431259−14.9
OpenWorldEngineer (Mean@8)Code39.024.1−14.9
NoCoT-Bench NCKI (kspine v3)General QA107128+14.7
FrontierCode 2.0 - ScoreCode42.227.6−14.6
SAFE-Teen 106: Health Advice (under-18 prompt)Safety16.930.8−13.8
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting41.554.7+13.2
DeepFinanceFinance & Accounting54.641.4−13.2
HealthBench (length-adjusted, Artificial Analysis)Health & Medicine62.149.1−13.0
Concept Synth INDUCTION Challenge100 Round 1General QA44.057.0+13.0
SAFE-Teen 106 worst branch: Health Advice (under-18 prompt)Safety30.843.6−12.8
Real-SWE 1.1Code51.239.2−12.0
DAYJOB: Data ScienceProfessional Work35.323.3−12.0
MineBench — Bradley-Terry RatingGames19982195+11.9
LMArena Code Arena WebDevCode17731682−11.8
SAFE-Teen 106: Relationships (no prompt)Safety16.928.5−11.6
AA-PersonalityBench Expressiveness indexChat & Writing73.384.8+11.5
Dual-Arm Sim (task completion)Agents0.011.3+11.3
HANDBOOK.md Agents v1.1Professional Work47.736.6−11.1
MedXpertQA MM Extended (Mercor held-out)Health & Medicine55.666.7+11.1
FrontierML ($200 wallet)Agents50.439.5−10.9
DualEntry Accounting AI BenchmarkFinance & Accounting80.291.1+10.9
Humanity's Last Exam DiamondGeneral QA53.342.5−10.8
ALE-BenchCode24482747+10.6
Agent Initiative BenchmarkAgents6.75.6−10.5
BuseyBench SVGMultimodal6.57.6+10.3
Bug Hunt BenchCode48.938.7−10.2
ChartographyMultimodal61.671.6+10.0
LMArena Agent Arena Net Improvement (Steering Burden)Agents16.010.7−9.8
SpeciEval (Pooled English Runs)Safety91.081.3−9.7
Humanity's Last ExamGeneral QA46.155.6+9.5
LisanBench Difficulty-WeightedGames29833849+9.3
GDP.xlsxProfessional Work29.138.3+9.2
SAFE-Teen 106: Urgent Safety (under-18 prompt)Safety2.912.1−9.2
LMArena Agent Arena Bash ReliabilityAgents12.7-5.4−9.0
TxBench Antibody Discovery V1 (100 evaluations)Science54.763.7+9.0
Toloka Arena (v3)Agents64.073.0+9.0
HardGeoBench <=100miMultimodal50.058.8+8.8
AA-AnalystAgent (pass^5)Professional Work57.548.8−8.7
LMArena Text Arena EloChat & Writing14721522+8.7
Agents' Last Exam v1.1 ScoreAgents60.469.0+8.6
ACCoRDGeneral QA85.777.3−8.4
AA-3D v1.0 EloCode12041147−8.3
AA-Briefcase v1.1 EloAgents16341485−8.2
o11y-bench 1.1 Pass^3Agents89.181.3−7.8
LAB-Bench ProtocolQA (SecureBio)Science88.395.9+7.6
DAYJOB: TaxFinance & Accounting10.83.2−7.6
SWE-Marathon v1.1Code51.343.8−7.5
FrontierSWE V2 Best@5Code71.764.3−7.4
ARC-AGI-2General QA83.991.3+7.4
BioSecBench-Function V1 Clean RerunsScience51.744.4−7.3
APEX-Accounting 1.1Finance & Accounting64.571.5+7.0
ATLAS (Exa, 200-task subset)Search46.853.8+7.0
WellnessBenchSafety57.050.0−7.0
BioSecBench-Structure V1 (58 evaluations)Science21.814.9−6.9
SAFE-Teen 106 worst branch: Relationships (under-18 prompt)Safety7.314.0−6.7
HealthBench ProfessionalHealth & Medicine70.864.2−6.6
FrontierSWE V2 Worst@5Code50.944.5−6.5
Access to Justice BenchmarkLegal74.968.5−6.5
ChessBench AccuracyGames75.881.8+6.4
Conceptual Reasoning IndexIndexes77.172.4−6.4
SpatialBench-Long V1 Clean RerunsScience34.928.6−6.4
VIALSMultimodal41.047.2+6.2
BullshitBench v2 LeaderboardSafety60.066.0+6.0
DistressBenchSafety74.668.6−5.9
VCT (SecureBio)Science52.558.4+5.9
MedCodeHealth & Medicine52.958.8+5.9
HealthBenchHealth & Medicine65.960.0−5.8
AA-PersonalityBench Conviction indexChat & Writing58.163.8+5.8
AA-PersonalityBench Warmth indexChat & Writing73.379.1+5.8
AutomationBench-AA 1.0.6Agents71.877.5+5.7
TxBench-Antibody-Development V1 (60 evaluations)Science36.731.1−5.6
GDPval Extended (Mercor held-out rubric score)Professional Work70.976.4+5.5
WCB (SecureBio)Science66.561.1−5.4
DAYJOB: FinanceFinance & Accounting24.018.8−5.2
VCT v2 (SecureBio)Science52.257.3+5.1
AA-Omniscience IndexIndexes32.342.4+5.0
LMCAGeneral QA68.363.7−4.6
Terminal-Bench 4.0Code61.857.4−4.4
MBCT (SecureBio)Science58.863.2+4.4
DAYJOB: HealthcareHealth & Medicine15.210.8−4.4
CritPt (avg@5)Science31.427.1−4.3
Artificial Analysis Intelligence Index v5.0.1Indexes54.550.4−4.1
τ³-Banking (Mercor run, bm25_grep retrieval)Agents54.058.1+4.1
ProgramBenchCode6.52.5−4.0
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine29.333.3+4.0
FrontierCode 1.1 Extended - ScoreCode64.460.6−3.9
Design Arena Overall Code CategoriesCode13491326−3.8
Physical-Engineering's Last Exam (hidden test)Professional Work44.941.1−3.8
SpatialBench V1 Clean RerunsScience72.869.0−3.8
FrontierCode 1.1 Extended - Pass RateCode69.966.1−3.7
APEX-Agents 1.1Professional Work83.787.4+3.7
MedScribeHealth & Medicine91.187.4−3.7
GDPval-AA v2.1Professional Work15511634+3.6
MysteryMechanismScience49.145.5−3.6
Artificial Analysis Intelligence Index v4.3.2Indexes56.052.6−3.4
SWE-Bench Pro V2 — FullCode95.899.2+3.4
AML BenchFinance & Accounting67.564.1−3.4
Decision-Bench v1.1Professional Work51.254.5+3.4
LMArena Agent Arena Praise vs ComplaintAgents30.524.1−3.2
LMArena Agent Arena Confirmed SuccessAgents19.513.2−3.1
ComplexConstraints (November 2026 rescoring)Chat & Writing67.270.3+3.1
Artificial Analysis Progress Index v1Indexes351342−3.1
GDP.pdfProfessional Work31.428.4−3.0
HPCT (SecureBio)Science77.274.2−3.0
LAB-Bench CloningScenarios (SecureBio)Science97.0100.0+3.0
CritPt v2 (agentic, avg@5)Science70.072.9+2.9
Terminal-Bench Science 0.1 (Vals)Science47.144.3−2.9
FrontierMath Tiers 1-3 (v2) - PrivateMath88.886.0−2.8
micro1 PharmaBenchHealth & Medicine11.48.6−2.8
CursorBench 4.0Code55.552.9−2.6
SWE Atlas v1.1 - OverallCode67.570.1+2.6
Public Benefits Bench v1.1Professional Work67.269.8+2.6
MMMU ProMultimodal85.888.3+2.5
Furniture Assembly BenchmarkMultimodal75.072.5−2.5
HardGeoBench <=10miMultimodal40.042.5+2.5
FrontierMath Tier 4 (v2) - PrivateMath80.578.0−2.4
BioSimBench 1.0Science16.018.4+2.4
RiemannBench (November 2026 rescoring)Math76.874.4−2.4
SWE-bench Multimodal (Vals v1, 480 Tasks)Code58.356.0−2.3
GameBench 2Games91.489.2−2.3
DAXBench v2Code86.288.3+2.1
ARC-AGI-1 Semi-PrivateGeneral QA96.598.5+2.0
PLE (Physical-Engineering's Last Exam)Professional Work45.243.2−2.0
AA-PersonalityBench Informativeness indexChat & Writing84.882.9−1.9
ScBench-Long V1 Clean RerunsScience30.328.4−1.9
SAGEMath51.853.6+1.9
Financial Coaching BenchFinance & Accounting74.973.1−1.8
HardGeoBench Avg DistanceMultimodal683741−1.8
GPQA-Bio (SecureBio)Science87.689.3+1.7
Code MigrationCode69.868.2−1.7
LiveBench 2.0Agents82.584.1+1.6
SAGE v1.1Math88.089.5+1.4
FrontierMath ErdősMath2.91.5−1.4
VariantBench V1 Clean RerunsScience51.750.3−1.4
SAFE-Teen 106: Relationships (under-18 prompt)Safety1.73.0−1.3
EpiBench V1 Clean RerunsScience30.228.9−1.3
ScBench V1 Clean RerunsScience60.359.1−1.2
PubMedQA (SecureBio)Health & Medicine81.382.4+1.1
MMLU-Bio (SecureBio)Science91.092.1+1.1
WMDP-Bio (SecureBio)Science92.691.5−1.1
Excel Modeling Benchmark v1.1Finance & Accounting75.974.8−1.1
Bio Capabilities Index (SecureBio)Indexes175174−1.0
Mystery Game PuzzlesGames65.064.0−1.0
ProofBench v1.1Math100.099.0−1.0
AA-LCR Extended (Mercor held-out)Search73.874.7+0.9
TxBench Oligo Discovery V1 (113 evaluations)Science51.951.0−0.9
Immigration BenchLegal69.068.1−0.9
GDPxProfessional Work8.27.3−0.9
SciCode 1.0.1 (Artificial Analysis)Code61.061.8+0.8
MKA (Artificial Analysis)Professional Work52.051.3−0.7
LMArena Agent Arena Steering BurdenAgents14.513.2−0.7
Political Science LLM BenchmarkProfessional Work68.969.6+0.6
GDPval (Mercor 213-task rubric score)Professional Work86.585.9−0.6
GBENCH One-shot Coding GScoreGames67.467.9+0.5
Excel Modeling BenchmarkFinance & Accounting75.775.2−0.5
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting85.786.0+0.3
TxBench Preclinical Pharmacology V1 Clean RerunsScience64.364.7+0.3
WeirdML v3Science19.920.3+0.3
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science17.317.6+0.3
MissionReportProfessional Work63.864.1+0.3
DTBenchGeneral QA97.397.6+0.3
LMArena Agent Arena Tool HallucinationAgents0.90.5−0.2
Epoch Capabilities Index (ECI)Indexes166167+0.2
Prophet Arena Arena Score (Fixed Context)Forecasting85.385.30.0
AA-LCR v1.1Search79.779.70.0
Chess Puzzles (task version 2.0)Games59.059.00.0
HardGeoBench Country AccuracyMultimodal62.562.50.0
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 - Data Privacy CybersecurityLegal20.020.00.0
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 - InsuranceLegal0.00.00.0
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
OTIS Mock AIME 2024-2025Math100.0100.00.0

Claude Sonnet 5.5vsGemini 4 Argonnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio 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…