Atlas

Compare Models

Claude Opus 5 vs Gemini 4 Argon

Gemini 4 Argon leads by 1.8 capability points (≈95% interval −0.6 to +4.3): statistically tied · P(higher) ≈ 93%

Wins 147 of 227 shared benchmarks (63 lost, 17 tied)

Capability vs. relative cost
Claude Opus 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 Opus 5vsGemini 4 Argoncapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins63 – 147 · 227 shared, 17 tied

Comparing matching setups only (36 benchmarks excluded) · suites counted once
← Claude Opus 5Gemini 4 Argon →

Claude Opus 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 Opus 5.

HPCT Refusal (CAIS)Safety100.029.0−71.0
ARC-AGI-3 (Semi-Private)General QA30.297.5+67.3
HieroglyphBenchMultimodal14.669.7+55.1
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
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal0.040.0+40.0
ChessBench EloGames8261485+39.2
CheatBenchSafety45.38.6+36.8
Machiavelli Deception (CAIS)Safety96.585.0+34.7
EQ-Bench 4Chat & Writing13651236−33.7
Time Horizon Index: KSPAgents18.852.5+33.7
LLM Creative Story WritingChat & Writing3.70.8−33.1
Agent Red Teaming k=100 (CAIS)Safety43.310.2+33.1
Concept Synth INDUCTION Challenge100 Round 1General QA24.057.0+33.0
Concept Synth INDUCTION Challenge64 Round 1General QA32.865.6+32.8
MASKSafety93.461.0−32.4
Terminal-Bench 2.1Code89.157.7−31.4
CrossSearch (pass^3)Search34.064.0+30.0
ChessBench CoherenceGames63.892.6+28.8
BrowseComp Extended (Mercor held-out)Search44.072.0+28.0
Humanity's Sixth SenseMultimodal30.555.4+24.9
LisanBench Path LengthGames154287560−24.4
BioSecBench-Refusal V1 Clean RerunsSafety24.548.7+24.2
Blueprint-Bench 2Multimodal0.30.5+24.0
TaxCalcBench TY25 (web search)Finance & Accounting50.074.0+24.0
TextQuests Game ProgressGames53.377.3+24.0
AutomationBench 1.0.6 (Zapier)Agents26.950.1+23.1
VIALSMultimodal24.447.2+22.8
AutomationBench-AA 1.0.6Agents56.677.5+20.9
Terminal-Bench 4.0 (Artificial Analysis)Code34.355.1+20.7
EnigmaEvalGeneral QA43.964.6+20.7
Terminal-Bench-Science 0.1Science28.648.6+20.0
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 - Data Privacy CybersecurityLegal0.020.0+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 - Structured Finance SecuritizationLegal20.00.0−20.0
ExploitGym - release-reported percentage, cohort unspecifiedCybersecurity22.142.0+19.9
NoCoT-Bench NCRI 15.2General QA115151+19.8
SnorkelLegalLegal26.446.0+19.6
Creative Writing v3 EloChat & Writing21351758−19.1
EBR-bench v4 (card ban, single agent)Games45.764.8+19.0
VCT-Refusal (CAIS)Safety29.410.7−18.7
SpatialViz-BenchMultimodal75.894.2+18.5
Eval Connections One-Shot with TrapsGames74.092.0+18.0
ClockBench AccuracyMultimodal65.583.3+17.8
LMArena Agent Arena Bash RecoveryAgents13.6-21.4−17.5
Terminal-Bench Science 0.1 (Vals)Science27.144.3+17.1
Chess Puzzles (task version 2.0)Games43.059.0+16.0
ComplexConstraintsChat & Writing37.353.3+16.0
IOI (Vals v2)Code84.3100.0+15.7
ERQAMultimodal70.886.5+15.7
TxBench Antibody Discovery V1 (100 evaluations)Science48.763.7+15.0
LAB-Bench ProtocolQA (SecureBio)Science81.195.9+14.8
HardGeoBench Avg DistanceMultimodal1218741+14.7
Humanity's Last Exam DiamondGeneral QA38.653.2+14.6
APEX-SWECode63.749.2−14.5
τ³-Banking (Mercor run, bm25_grep retrieval)Agents44.058.1+14.1
GDPval Extended (Mercor held-out rubric score)Professional Work62.976.4+13.5
LLM Chess EloGames12851609+13.3
LisanBench Difficulty-WeightedGames50683849−13.1
Bug Hunt BenchCode25.738.7+13.0
BioTIER RefusalSafety100.087.2−12.8
BuseyBench SVGMultimodal6.37.6+12.6
HardGeoBench <=100miMultimodal46.358.8+12.5
HardGeoBench <=10miMultimodal32.545.0+12.5
MindCube TinyMultimodal84.396.7+12.4
TextQuests Average Cumulative HarmSafety22.626.1−11.8
MetagenomicsBench V1Science54.042.3−11.7
Furniture Assembly BenchmarkMultimodal60.872.5+11.7
ChessBench AccuracyGames71.081.8+11.6
MineBench — Bradley-Terry RatingGames20082195+11.3
HackerRank-BenchCode38.649.7+11.1
Toloka Arena (v3)Agents62.073.0+11.0
Code MigrationCode57.568.2+10.7
APEX-Agents 1.1Professional Work77.087.4+10.4
SnorkelManufacturingProfessional Work9.419.5+10.1
HardGeoBench Country AccuracyMultimodal53.863.8+10.0
IntPhys 2Multimodal65.174.6+9.5
HealthBench ProfessionalHealth & Medicine73.464.2−9.2
τ³-Banking (Fireworks, banking_knowledge)Agents49.158.2+9.1
TxBench Oligo Discovery V1 (113 evaluations)Science42.551.0+8.6
GPQA DiamondGeneral QA87.996.4+8.5
BioSecBench-Surveillance V1 Clean RerunsScience51.559.8+8.3
MysteryMechanismScience37.445.5+8.1
RiemannBenchMath68.076.0+8.0
ALE-BenchCode25822805+7.9
WCB (SecureBio)Science68.861.1−7.7
LMArena Agent Arena Bash ReliabilityAgents9.8-5.4−7.6
DAYJOB: FinanceFinance & Accounting11.318.8+7.5
Vals Science IndexIndexes47.955.4+7.4
Vibe Code Bench 1–100Code28.521.2−7.3
Agents' Last Exam (ALE)Agents32.239.5+7.3
APEX-Accounting 1.1Finance & Accounting64.371.5+7.2
Public Benefits Bench v1.1Professional Work76.969.8−7.2
HealthBenchHealth & Medicine67.160.0−7.1
Senior SWE-BenchCode34.727.7−7.0
BullshitBench v2 LeaderboardSafety73.066.0−7.0
VCT (SecureBio)Science51.458.4+7.0
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine40.033.3−6.7
SpeciEval (Pooled English Runs)Safety87.981.3−6.7
LMArena Agent Arena Net Improvement (Steering Burden)Agents7.210.7+6.6
GPQA-Bio (SecureBio)Science82.789.3+6.6
Snorkel Digital Work IndexIndexes21.828.3+6.6
ACCoRDGeneral QA83.877.3−6.5
Harvey HLAB — Criteria Pass RateLegal87.794.0+6.3
CursorBench 4.0Code46.652.9+6.3
Snorkel WorkplaceAgentsProfessional Work14.921.2+6.3
NoCoT-Bench NCKI (kspine v3)General QA119128+6.2
Artificial Analysis Intelligence Index v5.0.1Indexes44.250.4+6.2
SnorkelRevOpsProfessional Work12.218.4+6.2
ABC-Bench liquid handling (SecureBio)Science94.0100.0+6.0
LMArena Agent Arena Praise vs ComplaintAgents12.224.1+6.0
LMArena Text Arena EloChat & Writing14901522+5.7
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting49.254.7+5.5
SciCode 1.0.1 (Artificial Analysis)Code56.461.8+5.4
ScarfBench Smoke ScoreCode69.875.2+5.4
Longform Creative WritingChat & Writing86.380.9−5.4
FrontierSWE V2 Worst@5Code39.144.5+5.3
HiL-Bench (Human-in-Loop Benchmark)Agents57.062.0+5.0
AA-AnalystAgent (pass^5)Professional Work53.848.8−5.0
Mystery Game PuzzlesGames59.064.0+5.0
SnorkelUnderwrite 2.0Finance & Accounting18.323.3+5.0
SWE-Marathon v1.1Code48.843.8−5.0
FrontierMath Tier 4 (v2) - PrivateMath73.278.0+4.9
MedCodeHealth & Medicine63.658.8−4.8
DeepSWE v1.1Code73.277.9+4.7
OSWorld 2.1 Offline, 500-step Partial ScoreAgents79.274.7−4.5
GDP.pdfProfessional Work24.028.4+4.4
HPCT (SecureBio)Science78.674.2−4.4
LMArena Agent Arena Steering BurdenAgents4.413.2+4.4
GameBench 2Games84.889.2+4.4
MMMU ProMultimodal84.088.3+4.3
SAGEMath49.453.6+4.2
CUA-benchAgents9.04.8−4.2
AA-Briefcase v1.1 EloAgents15581485−4.0
Arena Alignment IndexSafety75.679.4+3.9
EpiBench V1 Clean RerunsScience25.228.9+3.8
FrontierSWE V2 Best@5Code60.764.3+3.7
MedScribeHealth & Medicine91.087.4−3.6
Creative Writing v3 Rubric ScoreChat & Writing16.315.6−3.5
FrontierCode 1.1 Extended - Pass RateCode69.666.1−3.5
Terminal-Bench 4.0Code53.957.4+3.5
Bio Capabilities Index (SecureBio)Indexes178174−3.5
Conceptual Reasoning IndexIndexes74.772.4−3.1
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science14.617.6+3.0
RSI-Exam 0.1 (Full 88)Agents0.50.5+3.0
MBCT (SecureBio)Science60.263.2+3.0
BioMysteryBench (Vals Terminus 2)Science79.376.3−3.0
Snorkel Agentic Coding 2.0Code38.941.8+2.9
MacroscopeBenchCode69.972.8+2.9
Artificial Analysis Progress Index v1Indexes334342+2.8
AA-Omniscience IndexIndexes37.142.4+2.6
MedXpertQA MM Extended (Mercor held-out)Health & Medicine64.166.7+2.6
Vals RSI Index v1.1 (four tasks)Agents33.030.6−2.5
DAYJOB: HealthcareHealth & Medicine8.410.8+2.4
Tax Agent BenchFinance & Accounting76.779.0+2.4
AA-LCR v1.1Search82.079.7−2.3
Snorkel SWE-bench CLICode14.016.3+2.3
DAXBench v2Code86.288.3+2.1
LiveBench 2.0Agents82.184.1+2.0
PCT Political Inconsistency (CAIS)Safety36.738.7−2.0
CritPt (avg@5)Science29.127.1−2.0
BioSecBench-Function V1 Clean RerunsScience42.544.4+2.0
Bedside BenchHealth & Medicine94.692.7−1.9
Artificial Analysis Intelligence Index v4.3.2Indexes50.852.6+1.8
Excel Modeling BenchmarkFinance & Accounting73.675.2+1.7
TxBench Preclinical Pharmacology V1 Clean RerunsScience63.064.7+1.7
Epoch Capabilities Index (ECI)Indexes164167+1.7
DistressBenchSafety70.268.6−1.6
LMArena Agent Arena Confirmed SuccessAgents10.213.2+1.5
AutoResearchExam (hidden-test AUARC)Agents0.60.6+1.5
FrontierMath ErdősMath0.01.5+1.5
CritPt v2 (agentic, avg@5)Science71.472.9+1.4
SnorkelFinance 2.0Finance & Accounting16.515.1−1.4
LegalBenchLegal87.088.3+1.3
GDPval-AA v2.1Professional Work16051634+1.3
WeirdML v3Science19.020.3+1.3
FrontierPhysics v0.1Science9.58.3−1.2
LMArena Code Arena WebDevCode16911682−1.2
ScBench-Long V1 Clean RerunsScience27.328.4+1.1
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
micro1 PharmaBenchHealth & Medicine9.78.6−1.1
ARC-AGI-1 Semi-PrivateGeneral QA97.598.5+1.0
MMLU-Bio (SecureBio)Science91.192.1+1.0
OSWorld 2 (Aug 8) Offline, 500-step Partial ScoreAgents70.269.2−1.0
ScBench V1 Clean RerunsScience58.359.1+0.9
ARC-AGI-2General QA90.491.3+0.8
Design Arena Overall Code CategoriesCode13211326+0.8
AA-LCR Extended (Mercor held-out)Search73.974.7+0.8
Humanity's Last ExamGeneral QA56.455.6−0.8
LMCAGeneral QA64.563.7−0.8
FrontierCode 2.0 - Pass RateCode31.430.6−0.7
PLE (Physical-Engineering's Last Exam)Professional Work42.643.2+0.6
SpatialBench-Long V1 Clean RerunsScience28.028.6+0.6
BioSecBench-Structure V1 (58 evaluations)Science15.514.9−0.6
SpatialBench V1 Clean RerunsScience68.469.0+0.6
FrontierCode 2.0 - ScoreCode28.127.6−0.5
Legal Research BenchLegal55.354.8−0.5
Wiz Cyber Model Arena v3.9Cybersecurity71.270.8−0.4
Physical-Engineering's Last Exam (hidden test)Professional Work41.541.1−0.4
FrontierMath Tiers 1-3 (v2) - PrivateMath85.686.0+0.4
GDPval (Mercor 213-task rubric score)Professional Work85.685.9+0.3
VariantBench V1 Clean RerunsScience50.650.3−0.3
DTBenchGeneral QA97.997.6−0.3
GBENCH One-shot Coding GScoreGames68.167.9−0.3
LMArena Agent Arena Tool HallucinationAgents0.90.5−0.2
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 - Corporate M&ALegal20.020.00.0
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 - Healthcare Life SciencesLegal20.020.00.0
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 - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
LAB-Bench CloningScenarios (SecureBio)Science100.0100.00.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine13.913.90.0
Next.js AI Agent EvaluationsAgents100.0100.00.0
ProofBench v1.1Math99.099.00.0

Claude Opus 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…