Atlas

Compare Models

Gemini 4 Argon vs GPT-6 Sol

Gemini 4 Argon leads by 2.2 capability points (≈95% interval −0.3 to +4.6): statistically tied · P(higher) ≈ 96%

Wins 159 of 204 shared benchmarks (31 lost, 14 tied)

Capability vs. relative cost
Gemini 4 ArgonGPT-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.

Gemini 4 ArgonvsGPT-6 Solcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins159 – 31 · 204 shared, 14 tied

Comparing matching setups only (83 benchmarks excluded) · suites counted once
← Gemini 4 ArgonGPT-6 Sol →

Gemini 4 ArgonvsGPT-6 Solwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio and 40 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 Gemini 4 Argon.

ARC-AGI-3 (Semi-Private)General QA97.35.9−91.4
AA-PersonalityBench Informativeness indexChat & Writing82.910.5−73.1
AA-PersonalityBench Slop indexChat & Writing67.61.0−67.3
AA-PersonalityBench Expressiveness indexChat & Writing84.833.4−51.9
AA-PersonalityBench Organization indexChat & Writing56.212.4−44.2
HieroglyphBenchMultimodal69.726.0−43.7
Harvey's Legal Agent Benchmark - Capital MarketsLegal40.00.0−40.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal40.00.0−40.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal40.00.0−40.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal40.00.0−40.0
Terminal-Bench-Science 0.1Science48.68.6−40.0
Terminal-Bench 4.0 (Artificial Analysis)Code55.118.7−36.4
Terminal-Bench 2.1Code57.785.4+27.7
Tax Agent BenchFinance & Accounting79.053.0−26.0
Legal Research BenchLegal54.828.8−26.0
AA-PersonalityBench Warmth indexChat & Writing79.154.3−25.0
BrokenArXiv 08/2026 (Native Harnesses)Math54.278.6+24.4
Humanity's Sixth SenseMultimodal55.431.2−24.2
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting77.053.6−23.3
HardGeoBench Avg DistanceMultimodal7411497−23.3
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal27.44.1−23.3
Finance Agent v2 - Market AnalysisFinance & Accounting79.758.0−21.8
Finance Agent v2 - PrecedentsFinance & Accounting49.828.5−21.3
HardGeoBench <=100miMultimodal58.837.5−21.2
GDP.xlsxProfessional Work38.317.1−21.2
TxBench Antibody Discovery V1 (100 evaluations)Science63.743.3−20.3
NoCoT-Bench NCRI 15.2General QA151115−20.1
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal20.00.0−20.0
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 - ImmigrationLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Intellectual PropertyLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - International Trade SanctionsLegal20.00.0−20.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine13.933.9+20.0
Finance Agent v2 - All PassFinance & Accounting55.836.0−19.8
LegalReasonBenchLegal50.831.9−18.9
Finance Agent v2 - Disclosure AnalysisFinance & Accounting74.555.6−18.9
HardGeoBench <=10miMultimodal45.026.3−18.7
APEX-Agents 1.1Professional Work87.468.7−18.7
Creative Writing v3 EloChat & Writing17582126+18.6
Decision-Bench v1.1Professional Work54.536.5−18.0
ChartographyMultimodal71.653.6−18.0
GDPval Extended (Mercor held-out rubric score)Professional Work76.458.7−17.7
Blueprint-Bench 2Multimodal0.50.4−17.5
HardGeoBench Country AccuracyMultimodal63.846.3−17.5
IOI (Vals v2)Code100.082.6−17.4
AA-PersonalityBench Imagination indexChat & Writing46.729.6−17.3
AA-PersonalityBench Transparency indexChat & Writing21.939.1+17.3
Finance Agent v2 - AdjustmentsFinance & Accounting60.543.2−17.3
SpeciEval (Pooled English Runs)Safety81.398.5+17.2
AutomationBench 1.0.6 (Zapier)Agents50.133.2−16.9
ProofBench v1.1Math99.083.0−16.0
AutomationBench-AA 1.0.6Agents77.561.7−15.8
MysteryMechanismScience45.530.2−15.3
SimpleQA Verified (Epoch anti-abstention)General QA76.060.7−15.3
τ³-Banking (Mercor run, bm25_grep retrieval)Agents58.143.0−15.1
Toloka Arena (v3)Agents73.058.0−15.0
DualEntry Accounting AI BenchmarkFinance & Accounting91.176.2−14.9
LAB-Bench ProtocolQA (SecureBio)Science95.981.1−14.8
Terminal-Bench Science 0.1 (Vals)Science44.330.0−14.3
LMArena Agent Arena Bash RecoveryAgents-21.47.0+14.2
Furniture Assembly BenchmarkMultimodal72.558.3−14.2
SRE Bench — Fully SolvedCode44.330.5−13.7
AA-PersonalityBench Conviction indexChat & Writing63.850.5−13.5
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine33.320.0−13.3
Public Benefits Bench v1.1Professional Work69.856.6−13.1
Finance Agent v2 - Earnings AnalysisFinance & Accounting84.872.4−12.4
BioSecBench-Surveillance V1 Clean RerunsScience59.847.7−12.0
BullshitBench v2 LeaderboardSafety66.054.0−12.0
AA-Briefcase v1.1 EloAgents14851266−12.0
FrontierMath Tier 4 (v2) - PrivateMath78.090.0+12.0
APEX-Accounting 1.1Finance & Accounting71.559.7−11.8
HealthBenchHealth & Medicine60.048.2−11.8
MedCodeHealth & Medicine58.847.1−11.7
VariantBench V1 Clean RerunsScience50.338.7−11.6
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting81.570.0−11.5
LMArena Text Arena EloChat & Writing15221457−11.4
EBR-bench v4 (card ban, single agent)Games64.853.3−11.4
τ³-Banking (Fireworks, banking_knowledge)Agents58.247.0−11.3
RiemannBenchMath76.064.8−11.2
Code MigrationCode68.257.2−11.0
ChessBench EloGames14851303−10.8
Professional Reasoning Benchmark - LegalLegal49.438.6−10.8
BioSecBench-Refusal V1 Clean RerunsSafety48.737.9−10.8
Humanity's Last Exam (Text-Only)General QA57.146.3−10.8
Harvey HLAB — Criteria Pass RateLegal94.083.5−10.5
ArXivMath 08/2026 (Native Harnesses)Math82.593.0+10.5
Financial Coaching BenchFinance & Accounting73.162.6−10.5
Finance Agent v2 - Financial ModelingFinance & Accounting29.819.3−10.5
Vals Science IndexIndexes55.445.0−10.4
GDPval-AA v2.1Professional Work16341402−10.2
Finance Agent v2 - ComparablesFinance & Accounting51.040.8−10.1
RCM-Bench-1Health & Medicine91.581.5−10.0
FrontierCode 2.0 (Artificial Analysis)Code25.735.6+9.9
GameBench 2Games89.279.4−9.7
DAYJOB: FinanceFinance & Accounting18.89.3−9.5
SAGEMath53.644.8−8.9
BrowseComp Extended (Mercor held-out)Search72.063.2−8.8
Bio Capabilities Index (SecureBio)Indexes174161−8.7
Terminal-Bench 4.0Code57.649.4−8.2
Arena Alignment IndexSafety79.487.6+8.2
LMArena Agent Arena Bash ReliabilityAgents-5.410.9+8.2
ChessBenchGames70.062.0−8.0
CritPt v2 (agentic, avg@5)Science72.980.9+8.0
VCT v2 (SecureBio)Science57.349.3−8.0
Mystery Game PuzzlesGames64.056.0−8.0
TxBench Oligo Discovery V1 (113 evaluations)Science51.043.1−8.0
LisanBench Difficulty-WeightedGames38493118−7.8
AA-Omniscience IndexIndexes42.427.1−7.6
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting54.747.1−7.6
LMArena Agent Arena Net Improvement (Steering Burden)Agents10.76.9−7.2
DAYJOB: HealthcareHealth & Medicine10.83.6−7.2
LMArena Agent Arena Praise vs ComplaintAgents24.110.0−7.1
EpiBench V1 Clean RerunsScience28.922.0−6.9
NoCoT-Bench NCKI (kspine v3)General QA128119−6.7
HPCT (SecureBio)Science74.267.8−6.4
Artificial Analysis Intelligence Index v5.0.1Indexes50.444.2−6.3
WellnessBenchSafety50.056.0+6.0
ALE-BenchCode28052637−5.9
MMMU ProMultimodal88.382.5−5.8
Vectara Hallucination LeaderboardSafety12.16.5+5.6
Design Arena Overall Code CategoriesCode13261293−5.5
MedScribeHealth & Medicine87.482.0−5.4
BioSecBench-Function V1 Clean RerunsScience44.439.3−5.2
HealthBench (length-adjusted, Artificial Analysis)Health & Medicine49.154.2+5.1
ACCoRDGeneral QA77.382.4+5.1
Artificial Analysis Intelligence Index v4.3.2Indexes52.647.6−4.9
BioSimBench 1.0Science18.413.6−4.8
ChessBench AccuracyGames81.877.4−4.7
LMCAGeneral QA63.759.1−4.6
Bedside BenchHealth & Medicine92.788.1−4.6
AA-LCR Extended (Mercor held-out)Search74.770.2−4.5
VCT (SecureBio)Science58.454.0−4.4
ChessBench CoherenceGames92.688.2−4.4
LMArena Agent Arena Confirmed SuccessAgents13.24.5−4.4
APEX-SWECode49.245.0−4.2
SciCode 1.0.1 (Artificial Analysis)Code61.857.6−4.2
AA-LCR v1.1Search79.783.7+4.0
FrontierMath Tiers 1-3 (v2) - PrivateMath86.089.8+3.9
CritPt (avg@5)Science27.130.9+3.7
Excel Modeling BenchmarkFinance & Accounting75.271.5−3.7
HealthBench ProfessionalHealth & Medicine64.261.2−3.0
ARC-AGI-1 Semi-PrivateGeneral QA98.595.5−3.0
Eval Connections One-Shot with TrapsGames92.089.0−3.0
Chess Puzzles (task version 2.0)Games59.056.0−3.0
SpatialBench-Long V1 Clean RerunsScience28.631.6+3.0
Artificial Analysis Progress Index v1Indexes342334−2.7
Humanity's Last Exam DiamondGeneral QA42.545.2+2.7
LLM Chess EloGames16091545−2.6
ComplexConstraintsChat & Writing53.350.8−2.5
Creative Writing v3 Rubric ScoreChat & Writing15.616.1+2.5
GDPval (Mercor 213-task rubric score)Professional Work85.983.5−2.4
ScBench V1 Clean RerunsScience59.161.5+2.4
AA-3D v1.0 EloCode11471131−2.2
GBENCH One-shot Coding GScoreGames67.970.1+2.2
LiveBench 2.0Agents84.181.9−2.2
MBCT (SecureBio)Science63.261.1−2.1
LisanBench Path LengthGames75608222+2.1
Epoch Capabilities Index (ECI)Indexes167164−2.0
TaxCalcBench TY25 (web search)Finance & Accounting74.072.0−2.0
Longform Creative WritingChat & Writing80.982.8+1.9
Conceptual Reasoning IndexIndexes72.471.0−1.9
TxBench Preclinical Pharmacology V1 Clean RerunsScience64.763.0−1.7
ARC-AGI-2General QA91.389.6−1.7
LMArena Agent Arena Steering BurdenAgents13.210.1−1.5
LAB-Bench CloningScenarios (SecureBio)Science100.098.5−1.5
MacroscopeBenchCode72.874.3+1.5
PubMedQA (SecureBio)Health & Medicine82.480.9−1.5
Wiz Cyber Model Arena v3.9Cybersecurity70.869.3−1.5
WMDP-Bio (SecureBio)Science91.590.0−1.5
BioMysteryBench (Vals Terminus 2)Science76.374.8−1.5
SpatialBench V1 Clean RerunsScience69.067.5−1.4
MedXpertQA MM Extended (Mercor held-out)Health & Medicine66.768.0+1.3
MMLU-Bio (SecureBio)Science92.190.8−1.3
BioSecBench-Structure V1 (58 evaluations)Science14.913.8−1.2
DistressBenchSafety68.668.0−0.7
WCB (SecureBio)Science61.160.5−0.6
DeepSWE v1.1 (Fireworks, avg@3)Code67.867.3−0.6
ProgramBenchCode2.52.0−0.5
GDP.pdf (Artificial Analysis, All-pass)Professional Work25.625.2−0.4
GPQA-Bio (SecureBio)Science89.388.9−0.4
MetagenomicsBench V1Science42.342.0−0.3
BuseyBench SVGMultimodal7.67.5−0.3
Harvey LAB-AA v1.1 (hallucination-gated)Legal3.93.6−0.3
DTBenchGeneral QA97.697.3−0.3
LMArena Agent Arena Tool HallucinationAgents0.51.0+0.2
CyberBench v1.1 OverallCybersecurity77.978.0+0.1
FrontierMath ErdősMath1.51.50.0
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 - 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
ScBench-Long V1 Clean RerunsScience28.428.40.0

Gemini 4 ArgonvsGPT-6 Solnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio and 40 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…