Atlas

Compare Models

Gemini 4 Argon vs GPT-6 Astra

GPT-6 Astra leads by 1.3 capability points (≈95% interval −1.3 to +3.9): statistically tied · P(higher) ≈ 84%

Gemini 4 Argon wins 142 of 291 shared benchmarks (130 lost, 19 tied)

Capability vs. relative cost
Gemini 4 ArgonGPT-6 AstraClosed 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 Astracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins142 – 130 · 291 shared, 19 tied

Comparing matching setups only (144 benchmarks excluded) · suites counted once
← Gemini 4 ArgonGPT-6 Astra →

Gemini 4 ArgonvsGPT-6 Astrawins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio and 67 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6 Astra minus Gemini 4 Argon.

AA-PersonalityBench Transparency indexChat & Writing21.994.3+73.1
AA-PersonalityBench Informativeness indexChat & Writing82.929.6−53.8
AA-PersonalityBench Slop indexChat & Writing67.616.2−51.9
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
Time Horizon Index: KSPAgents52.590.5+38.0
MASKSafety61.098.5+37.5
Concept Synth INDUCTION Challenge100 Round 1General QA57.094.0+37.0
CyberBench v1.1 OverallCybersecurity77.941.1−36.8
Agent Initiative BenchmarkAgents5.69.0+33.7
Concept Synth INDUCTION Challenge64 Round 1General QA65.698.4+32.8
Terminal-Bench 2.1Code57.789.9+32.2
AA-PersonalityBench Imagination indexChat & Writing46.777.2+30.8
LLM Creative Story WritingChat & Writing0.83.4+29.9
HieroglyphBenchMultimodal69.740.0−29.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine13.941.7+27.8
DAYJOB: LawLegal9.936.9+27.0
AA-PersonalityBench Expressiveness indexChat & Writing84.858.1−26.9
Machiavelli Deception (CAIS)Safety85.076.3+26.3
SnorkelLegalLegal46.020.9−25.1
ABC-Bench liquid handling (SecureBio)Science100.076.7−23.3
Creative Writing v3 EloChat & Writing17582177+21.2
HPCT Refusal (CAIS)Safety29.050.0+21.0
OpenWorldEngineer (Mean@8)Code24.144.2+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 - 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
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal40.020.0−20.0
Mystery Game PuzzlesGames64.084.0+20.0
HLAB-Vals-1.1Legal27.17.1−20.0
FrontierMath Tier 4 (v2) - PrivateMath78.097.6+19.6
Finance Agent v2 - PrecedentsFinance & Accounting49.830.3−19.5
AA-PersonalityBench Conviction indexChat & Writing63.844.8−19.2
Terminal-Bench Science 0.1 (Vals)Science44.362.9+18.6
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal27.49.2−18.2
Visa Agentic Commerce BenchmarkAgents82.264.4−17.8
CheatBenchSafety28.845.0−16.2
ChessBench EloGames14851755+16.1
Tax Agent BenchFinance & Accounting79.063.3−15.7
Finance Agent v2 - All PassFinance & Accounting55.840.3−15.5
GDP.xlsxProfessional Work38.322.9−15.4
Legal Research BenchLegal54.839.4−15.4
Finance Agent v2 - Earnings AnalysisFinance & Accounting84.869.5−15.3
SpeciEval (Pooled English Runs)Safety81.396.5+15.3
Chess Puzzles (task version 2.0)Games59.074.0+15.0
sudo L7 (October 2026 rescoring)Code16.131.1+15.0
CloudOpsBench v1.0Agents57.542.8−14.8
FrontierCode 2.0 (Artificial Analysis)Code25.740.1+14.4
Finance Agent v2 - Market AnalysisFinance & Accounting79.765.5−14.2
WellnessBenchSafety50.064.0+14.0
Code2Video-AA EloCode12591427+13.6
LMArena Agent Arena Bash RecoveryAgents-21.45.5+13.5
AA-PersonalityBench Organization indexChat & Writing56.269.5+13.5
τ³-Banking (Fireworks, banking_knowledge)Agents58.244.8−13.4
Finance Agent v2 - ComparablesFinance & Accounting51.037.8−13.2
HoneyBench v0.2 (hack rate)Safety63.650.8+12.9
Finance Agent v2 - Disclosure AnalysisFinance & Accounting74.561.7−12.8
ALE-BenchCode28053164+12.7
SWE-bench Multimodal (Vals v1, 480 Tasks)Code56.043.3−12.7
SRE Bench — Fully SolvedCode44.356.9+12.6
WARC-Bench (192-task test split)Agents76.689.1+12.5
Ramp Accounting BenchFinance & Accounting61.148.7−12.4
APEX-Agents 1.1Professional Work87.475.1−12.3
GDPval Extended (Mercor held-out rubric score)Professional Work76.464.1−12.3
Finance Agent v2 - AdjustmentsFinance & Accounting60.548.3−12.2
MirrorCode (ML, +Private, 2L)Code58.946.7−12.2
DAYJOB: AccountingFinance & Accounting18.030.0+12.0
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine33.321.3−12.0
EBR-bench v4 (card ban, single agent)Games64.876.2+11.4
τ³-Banking (Mercor run, bm25_grep retrieval)Agents58.146.7−11.4
GBENCH One-shot Coding GScoreGames67.979.2+11.4
Legacy-Bench (Fireworks, avg@3)Code44.355.7+11.3
LegalReasonBenchLegal50.839.5−11.3
BioSecBench-Surveillance V1 Clean RerunsScience59.848.5−11.3
CritPt v2 (agentic, avg@5)Science72.983.7+10.9
SpatialBench-Long V1 Clean RerunsScience28.639.3+10.7
FrontierSWE V2 Worst@5Code44.555.1+10.7
RareDiseaseBench V1 (35 evaluations)Science43.833.3−10.5
TextQuests Average Cumulative HarmSafety26.129.2−10.5
Vals Finance IndexIndexes66.055.6−10.4
MedCodeHealth & Medicine58.848.5−10.3
DualEntry Accounting AI BenchmarkFinance & Accounting91.181.2−9.9
Code Migration v1.1Code70.880.6+9.8
Vals Science IndexIndexes55.465.1+9.7
SnorkelManufacturingProfessional Work19.510.0−9.5
HealthBench (length-adjusted, Artificial Analysis)Health & Medicine49.158.2+9.2
Senior SWE-BenchCode27.736.8+9.1
AutomationBench-AA 1.0.6Agents77.568.5−9.0
BioResearchBenchScience13.04.0−9.0
Web Clone BenchCode8.617.5+8.9
Dual-Arm Sim (task completion)Agents11.32.5−8.7
FrontierSWE V2 Best@5Code64.373.0+8.7
AutomationBench 1.0.6 (Zapier)Agents50.141.4−8.7
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting77.068.4−8.5
DAXBench v2Code88.396.8+8.5
LMArena Text Arena EloChat & Writing15221474−8.4
Vals Legal IndexIndexes76.868.4−8.4
Arena Alignment IndexSafety79.487.8+8.3
ClockBench AccuracyMultimodal83.375.0−8.3
Finance Agent v2 - Financial ModelingFinance & Accounting29.821.6−8.2
BioSecBench-Refusal V1 Clean RerunsSafety48.740.8−7.9
LisanBench Difficulty-WeightedGames38493113−7.9
FrontierMath Tiers 1-3 (v2) - PrivateMath86.093.7+7.7
TxBench Antibody Discovery V1 (100 evaluations)Science63.756.0−7.7
MysteryMechanismScience45.553.2+7.7
HardGeoBench <=100miMultimodal58.851.3−7.5
micro1 PharmaBenchHealth & Medicine8.616.1+7.5
Furniture Assembly BenchmarkMultimodal72.580.0+7.5
HardGeoBench <=10miMultimodal45.037.5−7.5
SnorkelRevOpsProfessional Work18.410.9−7.5
Humanity's Last Exam DiamondGeneral QA53.260.6+7.4
MetagenomicsBench V1Science42.349.7+7.3
BioTIER RefusalSafety87.279.9−7.3
SAGEMath53.646.4−7.3
Finance Agent v2.1Finance & Accounting67.760.4−7.2
PLE (Physical-Engineering's Last Exam)Professional Work41.148.3+7.2
LMArena Agent Arena Net Improvement (Steering Burden)Agents10.714.5+7.2
VCT-Refusal (CAIS)Safety10.717.6+6.9
Physical-Engineering's Last Exam (hidden test)Professional Work38.945.8+6.9
APEX-Accounting 1.1Finance & Accounting71.564.7−6.8
Snorkel WorkplaceAgentsProfessional Work21.214.4−6.8
LMArena Agent Arena Bash ReliabilityAgents-5.47.8+6.6
LAB-Bench ProtocolQA (SecureBio)Science95.989.3−6.6
GDP.pdf (Artificial Analysis, All-pass)Professional Work25.632.2+6.6
Harvey HLAB — Criteria Pass RateLegal94.087.5−6.6
Bio Capabilities Index (SecureBio)Indexes174164−6.6
Vibe Code Bench 1–100Code21.227.6+6.4
Artificial Analysis Intelligence Index v5.0.1Indexes50.456.9+6.4
BioSecBench-Structure V1 (58 evaluations)Science14.921.3+6.3
Westworld Finance Diligence BenchFinance & Accounting60.754.5−6.3
TaxCalcBench TY25 (web search)Finance & Accounting74.068.0−6.0
Toloka Arena (v3)Agents73.067.0−6.0
MedXpertQA MM Extended (Mercor held-out)Health & Medicine66.760.8−5.9
Snorkel Agentic Coding 2.0Code41.847.6+5.8
DistressBenchSafety68.674.3+5.7
Financial Coaching BenchFinance & Accounting73.167.5−5.6
Terminal-Bench 4.0 (Artificial Analysis)Code55.149.5−5.6
HealthBench ProfessionalHealth & Medicine64.269.7+5.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience64.759.3−5.3
SciCode 1.0.1 (Artificial Analysis)Code61.856.5−5.3
Snorkel Digital Work IndexIndexes28.323.0−5.3
HANDBOOK.md Agents v1.1Professional Work36.631.4−5.2
HPCT (SecureBio)Science74.269.0−5.2
MacroscopeBenchCode72.878.0+5.2
GameBench 2Games89.294.3+5.2
HardGeoBench Avg DistanceMultimodal741908−5.1
EnigmaEvalGeneral QA64.669.7+5.1
DataCenterBench Tier 1Agents35.040.0+5.0
ScBench V1 Clean RerunsScience59.163.9+4.8
Terminal-Bench-Science 0.1Science48.653.3+4.8
Blueprint-Bench 2Multimodal0.50.5−4.7
CritPt (avg@5)Science27.131.7+4.6
ComplexConstraintsChat & Writing53.357.7+4.4
ChessBench CoherenceGames92.697.0+4.4
ACCoRDGeneral QA77.381.7+4.4
NoCoT-Bench NCRI 15.2General QA151159+4.3
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science17.613.3−4.2
GDPval-AA v2.1Professional Work16341538−4.2
ERQAMultimodal86.582.3−4.2
BioSecBench-Function V1 Clean RerunsScience44.448.6+4.2
MindCube TinyMultimodal96.792.7−4.0
CrossSearch (pass^3)Search64.060.0−4.0
RiemannBenchMath76.072.0−4.0
FrontierCode 1.1 Extended - ScoreCode60.664.5+3.9
Snorkel SWE-bench CLICode16.312.4−3.9
AA-PersonalityBench Warmth indexChat & Writing79.175.3−3.8
DeepSWE v1.1Code77.974.1−3.8
LMArena Agent Arena Praise vs ComplaintAgents24.131.7+3.8
ARC-AGI-2General QA91.395.0+3.8
Remote Labor Index (RLI)Professional Work17.120.8+3.8
HardGeoBench Country AccuracyMultimodal63.860.0−3.7
VariantBench V1 Clean RerunsScience50.346.6−3.7
NeuroscienceBench V1 (91 evaluations)Science19.122.7+3.6
Build-A-Chip-Bench (20-task index run)Agents18.615.0−3.6
Excel Modeling BenchmarkFinance & Accounting75.271.7−3.5
SnorkelFinance 2.0Finance & Accounting15.111.6−3.5
ChessBench AccuracyGames81.885.0+3.4
Vectara Hallucination LeaderboardSafety12.18.7+3.4
HiL-Bench (Human-in-Loop Benchmark)Agents62.065.3+3.3
DrugDiscoveryBenchScience65.468.7+3.3
Agent Red Teaming k=100 (CAIS)Safety10.213.5−3.3
AA-3D v1.0 EloCode11471168+3.2
TextQuests Game ProgressGames77.374.2−3.1
HealthBenchHealth & Medicine60.056.9−3.1
Decision-Bench v1.1Professional Work54.551.5−3.1
Last Translation Benchmark (LTBv1-eval, blind)Chat & Writing40.944.0+3.1
BullshitBench v2 LeaderboardSafety66.069.0+3.0
ProgramBenchCode2.55.5+3.0
BioMysteryBench (Vals Terminus 2)Science76.379.3+3.0
IntPhys 2Multimodal74.677.4+2.8
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting81.578.8−2.7
Artificial Analysis Progress Index v1Indexes342350+2.7
DAYJOB: FinanceFinance & Accounting18.821.5+2.7
VIALSMultimodal47.249.9+2.7
ARC-AGI-3 (Semi-Private)General QA97.399.9+2.6
BioSimBench 1.0Science18.415.9−2.5
AA-AnalystAgent (pass^5)Professional Work48.851.3+2.5
BuseyBench SVGMultimodal7.67.3−2.4
Professional Reasoning Benchmark - FinanceFinance & Accounting49.947.5−2.4
SWE-Bench Pro V2 — FullCode99.296.9−2.3
MissionReportProfessional Work64.161.8−2.3
GDPxProfessional Work7.39.6+2.3
Excel Modeling Benchmark v1.1Finance & Accounting74.872.7−2.1
Epoch Capabilities Index (ECI)Indexes167169+2.1
SnorkelUnderwrite 2.0Finance & Accounting23.321.2−2.1
LiveBench 2.0Agents84.186.1+2.0
TissueMicroscopyBench V1 (49 evaluations)Science21.819.7−2.0
Conceptual Reasoning IndexIndexes72.473.9+2.0
Terminal-Bench 4.0Code57.659.6+2.0
Humanity's Last ExamGeneral QA55.653.6−2.0
DAYJOB: Data ScienceProfessional Work23.321.3−2.0
Eval Connections One-Shot with TrapsGames92.094.0+2.0
SEC-Bench Pro - release-reported cohortCybersecurity83.485.4+2.0
MMMU ProMultimodal88.386.4−1.9
Creative Writing v3 Rubric ScoreChat & Writing15.615.9+1.9
GPQA-Bio (SecureBio)Science89.387.4−1.9
Longform Creative WritingChat & Writing80.982.8+1.9
NoCoT-Bench NCKI (kspine v3)General QA128131+1.9
HackerRank-BenchCode45.443.5−1.9
Humanity's Sixth SenseMultimodal55.453.6−1.8
Political Science LLM BenchmarkProfessional Work69.667.9−1.7
TxBench-Antibody-Development V1 (60 evaluations)Science31.132.8+1.7
DAYJOB: TaxFinance & Accounting3.24.8+1.6
WCB (SecureBio)Science61.162.6+1.5
FrontierMath ErdősMath1.52.9+1.5
SpatialBench V1 Clean RerunsScience69.067.5−1.4
PCT Political Inconsistency (CAIS)Safety38.737.4+1.3
PubMedQA (SecureBio)Health & Medicine82.481.2−1.2
ScBench-Long V1 Clean RerunsScience28.427.3−1.1
SpatialViz-BenchMultimodal94.293.1−1.1
Professional Reasoning Benchmark - LegalLegal49.448.4−1.0
MKA (Artificial Analysis)Professional Work51.350.2−1.0
AA-Briefcase v1.1 EloAgents14851503+1.0
TxBench Oligo Discovery V1 (113 evaluations)Science51.051.9+0.9
LMArena Agent Arena Steering BurdenAgents13.214.8+0.8
DAYJOB: HealthcareHealth & Medicine10.811.6+0.8
GDPval (Mercor 213-task rubric score)Professional Work85.986.7+0.8
MMLU-Bio (SecureBio)Science92.191.3−0.8
Bedside BenchHealth & Medicine92.793.4+0.7
WMDP-Bio (SecureBio)Science91.590.8−0.7
MBCT (SecureBio)Science63.263.9+0.7
VCT v2 (SecureBio)Science57.358.0+0.7
SWE Atlas v1.1 - OverallCode70.169.4−0.7
AA-Omniscience IndexIndexes42.443.7+0.7
LMCAGeneral QA63.764.4+0.7
GPQA DiamondGeneral QA96.495.8−0.6
EpiBench V1 Clean RerunsScience28.928.3−0.6
APEX-SWECode49.248.6−0.6
ChartographyMultimodal71.671.0−0.6
Wiz Cyber Model Arena v3.9Cybersecurity70.870.2−0.6
Prophet Arena Arena Score (Fixed Context)Forecasting85.384.7−0.6
AutoResearchExam (hidden-test AUARC)Agents0.60.6+0.6
LisanBench Path LengthGames75607393−0.5
MedScribeHealth & Medicine87.487.9+0.5
Code MigrationCode68.267.7−0.4
SimpleQA Verified (Epoch anti-abstention)General QA76.075.6−0.4
LMArena Agent Arena Confirmed SuccessAgents13.214.0+0.4
Immigration BenchLegal68.167.7−0.4
AA-LCR v1.1Search79.780.0+0.3
DTBenchGeneral QA97.697.3−0.3
LMArena Agent Arena Tool HallucinationAgents0.51.0+0.2
LLM Chess EloGames16091614+0.2
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting86.086.2+0.1
Artificial Analysis Intelligence Index v4.3.2Indexes52.652.7+0.1
BigFinanceBench private holdout (final-answer accuracy)Finance & Accounting54.754.6−0.1
VCT (SecureBio)Science58.458.5+0.1
ARC-AGI-1 Semi-PrivateGeneral QA98.598.50.0
ChessBenchGames70.070.00.0
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 - Corporate M&ALegal20.020.00.0
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
IOI (Vals v2)Code100.0100.00.0
LAB-Bench CloningScenarios (SecureBio)Science100.0100.00.0
o11y-bench 1.1 Pass^3Agents87.587.50.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
ProofBench v1.1Math99.099.00.0

Gemini 4 ArgonvsGPT-6 Astranormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, benchmarks.bio and 67 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…