Atlas

Compare Models

Gemini 4 Argon vs GPT-5.6 Sol

Gemini 4 Argon leads by 3.1 capability points (≈95% interval +0.7 to +5.4): a clear lead over GPT-5.6 Sol · P(higher) ≈ 99%

Wins 170 of 217 shared benchmarks (33 lost, 14 tied)

Capability vs. relative cost
Gemini 4 ArgonGPT-5.6 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-5.6 Solcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins170 – 33 · 217 shared, 14 tied

Comparing matching setups only (78 benchmarks excluded) · suites counted once
← Gemini 4 ArgonGPT-5.6 Sol →

Gemini 4 ArgonvsGPT-5.6 Solwins by categoryData Nov 15, 2026Data: Vals AI, benchmarks.bio, Artificial Analysis and 52 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-5.6 Sol minus Gemini 4 Argon.

ARC-AGI-3 (Semi-Private)General QA49.80.3−49.5
CheatBenchSafety28.874.9−46.1
HieroglyphBenchMultimodal69.727.6−42.1
Terminal-Bench 4.0 (Artificial Analysis)Code55.114.6−40.4
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
BrowseComp Extended (Mercor held-out)Search72.032.3−39.7
MirrorCode (ML, +Private, 2L)Code58.920.0−38.9
MASKSafety61.095.4+34.4
Terminal-Bench 2.1Code57.789.5+31.8
RareDiseaseBench V1 (35 evaluations)Science43.813.3−30.5
Agent Red Teaming k=100 (CAIS)Safety10.240.3−30.1
HiL-Bench (Human-in-Loop Benchmark)Agents62.032.3−29.7
Time Horizon Index: KSPAgents52.523.8−28.7
ChartographyMultimodal71.645.0−26.6
TxBench Antibody Discovery V1 (100 evaluations)Science63.737.3−26.3
LisanBench Difficulty-WeightedGames38491403−26.2
Terminal-Bench-Science 0.1Science48.622.4−26.2
CrossSearch (pass^3)Search64.038.0−26.0
Humanity's Sixth SenseMultimodal55.430.0−25.4
TextQuests Game ProgressGames77.352.1−25.2
Terminal-Bench Science 0.1 (Vals)Science44.320.0−24.3
Finance Agent v2 - PrecedentsFinance & Accounting49.826.2−23.6
NoCoT-Bench NCRI 15.2General QA151109−22.9
τ³-Banking (Fireworks, banking_knowledge)Agents58.235.7−22.6
FrontierSWE V2 Worst@5Code44.522.2−22.2
Humanity's Last Exam DiamondGeneral QA53.231.2−22.0
AutomationBench 1.0.6 (Zapier)Agents50.128.8−21.3
HardGeoBench Avg DistanceMultimodal7411429−21.2
Blueprint-Bench 2Multimodal0.50.3−20.8
VIALSMultimodal47.226.5−20.7
FrontierSWE V2 Best@5Code64.344.2−20.1
EBR-bench v4 (card ban, single agent)Games64.844.8−20.0
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 - 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
MindCube TinyMultimodal96.776.8−19.9
TissueMicroscopyBench V1 (49 evaluations)Science21.82.0−19.7
Terminal-Bench 4.0Code57.637.9−19.7
LLM Creative Story WritingChat & Writing0.82.5+18.9
BullshitBench v2 LeaderboardSafety66.048.0−18.0
TextQuests Average Cumulative HarmSafety26.131.4−17.8
SpeciEval (Pooled English Runs)Safety81.399.0+17.8
HardGeoBench <=100miMultimodal58.841.3−17.5
AutomationBench-AA 1.0.6Agents77.560.1−17.4
HardGeoBench Country AccuracyMultimodal63.847.5−16.2
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine13.930.0+16.1
ProofBench v1.1Math99.083.0−16.0
ChessBench EloGames14851218−15.9
Furniture Assembly BenchmarkMultimodal72.556.7−15.8
Finance Agent v2 - Market AnalysisFinance & Accounting79.764.2−15.5
BioTIER RefusalSafety87.271.9−15.3
ERQAMultimodal86.571.2−15.3
Code MigrationCode68.252.9−15.3
ClockBench AccuracyMultimodal83.368.3−15.0
HardGeoBench <=10miMultimodal45.030.0−15.0
VariantBench V1 Clean RerunsScience50.335.3−15.0
MedCodeHealth & Medicine58.844.0−14.8
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine33.318.7−14.7
BuseyBench SVGMultimodal7.66.1−14.5
Finance Agent v2 - All PassFinance & Accounting55.841.4−14.4
LAB-Bench ProtocolQA (SecureBio)Science95.981.9−14.0
Concept Synth INDUCTION Challenge100 Round 1General QA57.043.0−14.0
ALE-BenchCode28052410−14.0
Vals Science IndexIndexes55.441.5−13.9
TxBench Oligo Discovery V1 (113 evaluations)Science51.037.2−13.9
SRE Bench — Fully SolvedCode44.330.5−13.7
SpatialViz-BenchMultimodal94.280.7−13.6
Harvey HLAB — Criteria Pass RateLegal94.080.5−13.5
DAYJOB: FinanceFinance & Accounting18.85.5−13.3
BioSecBench-Surveillance V1 Clean RerunsScience59.847.3−12.5
TxBench-Antibody-Development V1 (60 evaluations)Science31.118.9−12.2
MysteryMechanismScience45.533.3−12.2
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting77.064.9−12.0
Toloka Arena (v3)Agents73.061.0−12.0
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science17.65.8−11.8
LMArena Agent Arena Net Improvement (Steering Burden)Agents10.74.6−11.5
Finance Agent v2 - AdjustmentsFinance & Accounting60.549.2−11.3
CursorBench 4.0Code52.941.7−11.2
Finance Agent v2 - Disclosure AnalysisFinance & Accounting74.563.3−11.2
Tax Agent BenchFinance & Accounting79.068.0−11.1
LMArena Agent Arena Bash RecoveryAgents-21.40.7+11.1
MetagenomicsBench V1Science42.331.3−11.0
Creative Writing v3 EloChat & Writing17581974+10.9
IntPhys 2Multimodal74.663.9−10.7
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting54.744.2−10.5
Finance Agent v2 - Earnings AnalysisFinance & Accounting84.874.4−10.5
Wiz Cyber Model Arena v3.9Cybersecurity77.266.9−10.3
AA-Omniscience IndexIndexes42.422.0−10.2
Humanity's Last ExamGeneral QA55.645.5−10.1
DAYJOB: HealthcareHealth & Medicine10.81.6−9.2
GameBench 2Games89.280.1−9.1
Finance Agent v2 - Financial ModelingFinance & Accounting29.820.7−9.1
Eval Connections One-Shot with TrapsGames92.083.0−9.0
SpatialBench-Long V1 Clean RerunsScience28.637.5+8.9
BioSecBench-Function V1 Clean RerunsScience44.435.5−8.9
DrugDiscoveryBenchScience65.456.5−8.9
MCP Atlas April 2026 (all 1,000 tasks)Agents90.781.8−8.9
LMArena Agent Arena Praise vs ComplaintAgents24.16.3−8.9
LisanBench Path LengthGames75604699−8.9
IOI (Vals v2)Code100.091.2−8.8
LMArena Agent Arena Bash ReliabilityAgents-5.411.4+8.4
ExploitGym - release-reported percentage, cohort unspecifiedCybersecurity42.033.7−8.3
AutoResearchExam (hidden-test AUARC)Agents0.60.5−8.1
VCT v2 (SecureBio)Science57.349.4−7.9
LiveBench 2.0Agents84.176.3−7.8
Concept Synth INDUCTION Challenge64 Round 1General QA65.657.8−7.8
DeepSWE v1.1Code77.970.7−7.2
Senior SWE-BenchCode27.734.7+7.0
WellnessBenchSafety50.057.0+7.0
Machiavelli Deception (CAIS)Safety85.082.7+7.0
HackerRank-BenchCode45.438.4−6.9
Legal Research BenchLegal54.848.1−6.7
LMArena Text Arena EloChat & Writing15221485−6.5
GPQA DiamondGeneral QA96.489.9−6.5
AA-Briefcase v1.1 EloAgents14851369−6.3
Mystery Game PuzzlesGames64.058.0−6.0
TaxCalcBench TY25 (web search)Finance & Accounting74.068.0−6.0
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting81.575.6−5.9
ChessBench AccuracyGames81.876.4−5.8
BioSecBench-Structure V1 (58 evaluations)Science14.920.7+5.8
Finance Agent v2 - ComparablesFinance & Accounting51.045.3−5.7
MMMU ProMultimodal88.382.7−5.7
Artificial Analysis Intelligence Index v4.3.2Indexes52.647.0−5.6
LMArena Agent Arena Confirmed SuccessAgents13.22.3−5.5
GDPval-AA v2.1Professional Work16341511−5.4
NoCoT-Bench NCKI (kspine v3)General QA128121−5.4
LMCAGeneral QA63.758.5−5.2
BioMysteryBench (Vals Terminus 2)Science76.371.1−5.2
OSWorld 2 (Aug 8) Offline, 500-step Partial ScoreAgents69.264.1−5.1
Conceptual Reasoning IndexIndexes72.468.7−5.1
MacroscopeBenchCode72.877.8+5.0
HPCT Refusal (CAIS)Safety29.024.0−5.0
DistressBenchSafety68.673.6+4.9
FrontierMath Tier 4 (v2) - PrivateMath78.082.9+4.9
Arena Alignment IndexSafety79.484.2+4.7
HealthBenchHealth & Medicine60.055.3−4.7
AA-LCR v1.1Search79.784.0+4.3
SEC-Bench Pro - release-reported cohortCybersecurity83.479.1−4.3
CritPt v2 (agentic, avg@5)Science72.977.1+4.3
VCT (SecureBio)Science58.454.3−4.1
SciCode 1.0.1 (Artificial Analysis)Code61.857.8−4.1
GPQA-Bio (SecureBio)Science89.385.3−4.0
LMArena Agent Arena Steering BurdenAgents13.25.5−3.8
SWE-Bench Pro V2 — FullCode99.295.5−3.7
Artificial Analysis Progress Index v1Indexes342331−3.6
ChessBench CoherenceGames92.689.1−3.5
NeuroscienceBench V1 (91 evaluations)Science19.115.8−3.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience64.761.3−3.3
WCB (SecureBio)Science61.164.4+3.3
Bio Capabilities Index (SecureBio)Indexes174169−3.3
Public Benefits Bench v1.1Professional Work69.866.5−3.2
FrontierMath Tiers 1-3 (v2) - PrivateMath86.089.1+3.2
EpiBench V1 Clean RerunsScience28.925.8−3.1
Creative Writing v3 Rubric ScoreChat & Writing15.616.2+3.1
PCT Political Inconsistency (CAIS)Safety38.741.7−3.0
Excel Modeling BenchmarkFinance & Accounting75.272.3−2.9
Epoch Capabilities Index (ECI)Indexes167163−2.8
ComplexConstraintsChat & Writing53.350.5−2.8
LLM Chess EloGames16091550−2.5
GDP.pdfProfessional Work28.430.7+2.3
MedScribeHealth & Medicine87.485.2−2.2
MBCT (SecureBio)Science63.261.1−2.1
HPCT (SecureBio)Science74.276.3+2.1
DTBenchGeneral QA97.695.7−1.9
WMDP-Bio (SecureBio)Science91.589.7−1.8
Financial Coaching BenchFinance & Accounting73.174.9+1.8
Bedside BenchHealth & Medicine92.791.1−1.6
RiemannBenchMath76.074.4−1.6
CyberBench v1.1 OverallCybersecurity77.976.3−1.5
FrontierMath ErdősMath1.50.0−1.5
CritPt (avg@5)Science27.128.6+1.4
ACCoRDGeneral QA77.375.9−1.4
LegalBenchLegal88.387.0−1.3
PubMedQA (SecureBio)Health & Medicine82.481.1−1.3
ABC-Bench liquid handling (SecureBio)Science100.098.7−1.3
AA-AnalystAgent (pass^5)Professional Work48.847.5−1.2
ARC-AGI-2General QA91.392.5+1.3
APEX-SWECode49.248.0−1.2
LAB-Bench CloningScenarios (SecureBio)Science100.098.8−1.2
Vibe Code Bench 1–100Code21.220.0−1.2
ScBench-Long V1 Clean RerunsScience28.429.6+1.1
SAGEMath53.652.6−1.1
ARC-AGI-1 Semi-PrivateGeneral QA98.597.5−1.0
MMLU-Bio (SecureBio)Science92.191.1−1.0
ProgramBenchCode2.51.5−1.0
SpatialBench V1 Clean RerunsScience69.068.1−0.9
Longform Creative WritingChat & Writing80.981.7+0.8
ScBench V1 Clean RerunsScience59.159.8+0.7
VCT-Refusal (CAIS)Safety10.711.3+0.6
Prophet Arena Arena Score (Fixed Context)Forecasting85.384.8−0.5
Vectara Hallucination LeaderboardSafety12.112.4−0.3
RCM-Bench-1Health & Medicine91.591.2−0.3
LMArena Agent Arena Tool HallucinationAgents0.51.0+0.2
GBENCH One-shot Coding GScoreGames67.967.7−0.2
FrontierCode 1.1 Extended - ScoreCode60.660.60.0
Chess Puzzles (task version 2.0)Games59.059.00.0
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 - 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 - 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

Gemini 4 ArgonvsGPT-5.6 Solnormalised score marginsData Nov 15, 2026Data: Vals AI, benchmarks.bio, Artificial Analysis and 52 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…