Atlas

Compare Models

Gemini 3.8 Flash vs Grok 4.6

Gemini 3.8 Flash leads by 0.7 capability points (≈95% interval −1.4 to +2.8): statistically tied · P(higher) ≈ 74%

Wins 132 of 281 shared benchmarks (131 lost, 18 tied)

Capability vs. relative cost
Gemini 3.8 FlashGrok 4.6Closed 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 3.8 FlashvsGrok 4.6capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins132 – 131 · 281 shared, 18 tied

Comparing matching setups only (31 benchmarks excluded) · suites counted once
← Gemini 3.8 FlashGrok 4.6 →

Gemini 3.8 FlashvsGrok 4.6wins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 69 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Grok 4.6 minus Gemini 3.8 Flash.

RT-SAFE Collisions per EpisodeSafety27.759.5−79.7
AA-PersonalityBench Organization indexChat & Writing73.314.3−59.6
Design Creativity BenchCode60.643.4−54.6
AA-PersonalityBench Transparency indexChat & Writing23.871.4+48.1
ClockBench AccuracyMultimodal66.720.0−46.7
Gray Swan IPI Attack Success (k=15)Safety5.551.8−46.3
LLM Chess EloGames1546443−45.3
Antidote: Everyday EditionSafety1087966−43.2
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal40.00.0−40.0
Document ArenaProfessional Work14821438−39.2
ChessBench EloGames1527891−37.8
AA-PersonalityBench Slop indexChat & Writing71.435.3−36.5
AA-PersonalityBench Informativeness indexChat & Writing81.044.8−36.5
HieroglyphBenchMultimodal54.620.3−34.3
LLM Creative Story WritingChat & Writing0.1-2.9−34.2
ARC-AGI-3 (Semi-Private)General QA35.02.1−32.9
Hemingway-benchChat & Writing1093983−30.5
ChessBenchGames73.343.3−30.0
AA-PersonalityBench Warmth indexChat & Writing60.031.5−28.8
ChessBench CoherenceGames94.669.2−25.4
ScBench V1 Clean RerunsScience34.059.1+25.1
Chess Puzzles (task version 2.0)Games58.033.0−25.0
CadQueryEval (Corrected Geometry)Code96.072.0−24.0
SWE-bench Verified Extended (Mercor held-out)Code54.078.0+24.0
Vending-Bench 2Games50949047+23.9
CyberBench v1.1 OverallCybersecurity43.867.7+23.9
BlenderWorldCode34.257.0+22.8
LAB-Bench LitQA2 (SecureBio)Science51.528.9−22.6
ARC-AGI-2General QA89.267.1−22.1
Concept Synth INDUCTION Challenge64 Round 1General QA14.135.9+21.9
Hedge-BenchFinance & Accounting4.726.4+21.7
Olam Social Poker EloGames14791499+21.7
NoCoT-Bench NCRI 15.2General QA12686.8−21.4
AA-PersonalityBench Imagination indexChat & Writing41.020.0−21.2
Shroombench (mushroom identification, top-1)Multimodal65.044.0−21.0
Roboflow Vision Evals (October 2026, five tasks)Multimodal79.859.1−20.7
SimpleQA Verified (Epoch anti-abstention)General QA69.749.3−20.4
ChartographyMultimodal40.920.5−20.4
BuseyBench SVGMultimodal6.64.6−20.2
BrowseComp Extended (Mercor held-out)Search41.521.5−20.0
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Capital MarketsLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal20.040.0+20.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Litigation Dispute ResolutionLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal20.00.0−20.0
APEX-SWECode36.456.4+20.0
SnorkelFinance 2.0Finance & Accounting5.425.1+19.7
AA-Briefcase EloAgents12021534+19.3
WeirdML (v2)Science84.867.3−17.5
Extended NYT Connections 940 (Quadratic v1)Games97.480.0−17.4
Epoch Automation ReportsAgents42.359.1+16.8
HANDBOOK.mdProfessional Work10.827.3+16.5
Biopharma Bench V0.1Health & Medicine46.663.0+16.4
Roboflow Vision Evals (September 2026 protocol)Multimodal85.168.7−16.4
Finance Agent v2 - Market AnalysisFinance & Accounting78.562.2−16.3
Concept Synth INDUCTION Challenge100 Round 1General QA9.025.0+16.0
LAB-Bench ProtocolQA (SecureBio)Science84.068.1−15.9
SWE-bench VerifiedCode80.095.6+15.6
SnorkelLegalLegal25.340.7+15.4
AA-PersonalityBench Expressiveness indexChat & Writing79.163.8−15.4
Humanity's Last Exam DiamondGeneral QA34.319.6−14.7
ChessBench AccuracyGames81.167.8−14.3
Finance Agent v2 - PrecedentsFinance & Accounting36.121.9−14.2
Finance Agent v2 - Financial ModelingFinance & Accounting32.618.9−13.7
Mystery Game PuzzlesGames47.034.0−13.0
Time Horizon Index: KSPAgents18.85.8−13.0
LLM Debate BenchmarkChat & Writing15141604+12.8
ALE-BenchCode21821822−12.7
MedXpertQA MM Extended (Mercor held-out)Health & Medicine60.848.4−12.4
Bio Capabilities Index (SecureBio)Indexes153137−12.0
MetagenomicsBench V1Science26.738.7+12.0
MetagenomicsBench V0Science26.738.7+12.0
ComplexConstraintsChat & Writing48.436.5−11.9
SciUniverse Level 1Science14.626.2+11.6
TxBench-Antibody-Development V1 (60 evaluations)Science15.026.1+11.1
ARC-AGI-1 Semi-PrivateGeneral QA98.587.5−11.0
Finance Agent v2 - ComparablesFinance & Accounting52.041.2−10.8
HealthBench ProfessionalHealth & Medicine59.048.5−10.5
SnorkelRevOpsProfessional Work5.415.8+10.4
MMMU-Pro Extended (Mercor held-out)Multimodal36.426.0−10.4
TxBench Oligo Discovery V1 Clean RerunsScience33.143.3+10.3
Bug Hunt BenchCode17.127.3+10.2
DAYJOB: FinanceFinance & Accounting2.812.8+10.0
BioMysteryBench (Vals Terminus 2)Science62.272.2+10.0
FrontierMath Tier 4 (v2) - PrivateMath22.031.7+9.8
SWE Atlas - Codebase QnACode40.650.3+9.7
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting68.177.8+9.7
AA-LCR Extended (Mercor held-out)Search55.665.2+9.6
MedXpertQA MM (200-question set)Health & Medicine84.775.2−9.5
Excel Modeling BenchmarkFinance & Accounting72.262.7−9.5
Medical Long Context Reasoning (MLCR-AA)Health & Medicine21.712.2−9.4
IOI (Vals v2)Code56.947.6−9.3
SnorkelManufacturingProfessional Work7.116.4+9.3
Legal Research BenchLegal38.948.1+9.1
VariantBench V1 Clean RerunsScience31.440.4+9.0
HLE Extended (Mercor held-out)General QA36.845.8+9.0
WordleBenchGames98.089.0−9.0
VIALSMultimodal29.420.5−8.9
GDPval-AA v2Professional Work14641643+8.8
LMArena Agent Arena Net ImprovementAgents3.81.1−8.7
SM BenchSafety80.489.0+8.6
DeepSWE v1.1Code73.765.2−8.5
Furniture Assembly BenchmarkMultimodal31.740.0+8.3
GDPval-AA v2.1Professional Work14421627+8.2
BioSecBench-Function V1 Clean RerunsScience36.044.0+8.1
Code MigrationCode36.544.6+8.0
FrontierCode 1.1 Extended - ScoreCode53.561.3+7.9
SpatialBench V1 Clean RerunsScience59.467.3+7.8
AA-PersonalityBench Conviction indexChat & Writing84.892.4+7.7
Vision Arena OverallMultimodal12931263−7.7
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting80.973.2−7.6
LiveBench 2.0Agents67.374.9+7.6
Finance Agent v2 - Earnings AnalysisFinance & Accounting81.473.9−7.5
FrontierCode 1.1 Extended - Pass RateCode59.667.0+7.4
CVDP (517-problem set)Code50.758.0+7.3
RedlineBenchLegal41.348.5+7.2
LMArena Text Arena EloChat & Writing14971456−7.2
EnterpriseBench: CoreCraftProfessional Work58.565.6+7.1
TxBench Preclinical Pharmacology V1 Clean RerunsScience54.761.7+7.0
SWE-bench MultilingualCode85.892.7+6.9
micro1 PharmaBenchHealth & Medicine5.612.5+6.9
Realm Financial ReasoningFinance & Accounting51.544.6−6.9
SUPERChemScience72.665.7−6.9
WCB (SecureBio)Science57.150.3−6.8
Finance Agent v2 - All PassFinance & Accounting49.742.9−6.8
ACCoRDGeneral QA78.271.6−6.5
SimpleBenchGeneral QA82.475.9−6.5
CAD ArenaAgents41.034.8−6.3
Real-SWECode38.832.5−6.2
NoCoT-Bench NCKI (kspine v3)General QA121112−6.2
SAGEMath35.128.9−6.2
AutomationBench-AA 1.0.6Agents60.967.0+6.1
CrossSearch (pass^3)Search29.035.0+6.0
GDP.pdfProfessional Work23.217.2−6.0
Snorkel WorkplaceAgentsProfessional Work11.217.1+5.9
MysteryMechanismScience36.530.6−5.9
LAB-Bench CloningScenarios (SecureBio)Science100.094.2−5.8
FrontierSWE V2Code19.625.3+5.7
Vals RSI Index v1.1 (four tasks)Agents19.925.5+5.5
Blueprint-Bench 2Multimodal0.40.3−5.4
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine13.318.7+5.3
Finance Agent v2 - AdjustmentsFinance & Accounting54.849.6−5.2
Artificial Analysis Coding Agent Index v1.5Indexes41.947.0+5.1
AA-AnalystAgent (pass^5)Professional Work46.341.3−5.0
𝜏³-BankingAgents45.850.7+4.9
Arena Alignment IndexSafety75.880.8+4.9
RuneBenchGames6.06.3+4.9
Humanity's Last ExamGeneral QA44.539.7−4.8
NeuroscienceBench V1 (91 evaluations)Science15.820.5+4.8
GameBench 2Games64.068.7+4.7
LMArena Agent Arena Net Improvement (Steering Burden)Agents2.3-0.2−4.7
LMArena Code Arena WebDevCode15831618+4.6
Tax Agent BenchFinance & Accounting66.871.4+4.6
LMCAGeneral QA52.948.5−4.5
APEX-Agents 1.1Professional Work72.977.3+4.4
Conceptual Reasoning IndexIndexes65.762.5−4.4
Access to Justice BenchmarkLegal67.171.3+4.3
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science7.311.5+4.2
Finance Agent v2 - Disclosure AnalysisFinance & Accounting68.564.4−4.1
MMMU ProMultimodal84.580.5−4.0
Veracode GenAI Code Security (Fall 2026)Cybersecurity62.058.0−4.0
BabaIsBench Stage 0: The IntroGames100.096.0−4.0
GeneBench-Pro Public Release SubsetScience11.015.0+4.0
WellnessBenchSafety47.043.0−4.0
Vals RSI Index v1.1 (historical five tasks)Agents17.621.5+4.0
Terminal-Bench-Science 0.1Science10.06.2−3.8
LMArena Agent Arena Praise vs ComplaintAgents4.9-2.5−3.7
Harvey LAB: In-House Contracting (held-out)Legal6.310.0+3.7
SWE-TogetherCode64.260.6−3.6
Snorkel Digital Work IndexIndexes16.920.5+3.6
MedCodeHealth & Medicine48.144.7−3.4
BioSecBench-Surveillance V1 Clean RerunsScience44.147.5+3.4
ScBench-Long V1 Clean RerunsScience19.916.5−3.4
BioSecBench-Surveillance V0 (102 tasks)Science44.147.5+3.4
Artificial Analysis Intelligence Index v4.3.2Indexes40.944.3+3.4
TaxEval v2Finance & Accounting74.471.1−3.4
AIME 2025Math96.7100.0+3.3
GBENCH One-shot Coding GScoreGames58.655.4−3.3
Artificial Analysis Intelligence Index v4.3Indexes41.244.4+3.2
Bedside BenchHealth & Medicine86.583.3−3.2
LMArena Agent Arena Bash RecoveryAgents-0.75.7+3.2
CloudOpsBench v1.0Agents42.839.7−3.1
Vals Index v2Indexes62.359.2−3.1
AA-LCR v1.1Search84.081.0−3.0
Eval Connections One-Shot with TrapsGames88.091.0+3.0
ProofBench v1.1Math48.051.0+3.0
Toloka Arena (v3)Agents59.056.0−3.0
Arena Image-to-WebDevCode15941607+2.9
SpatialBench V0 VerifiedScience67.069.9+2.9
SkillsBench without SkillsAgents41.038.2−2.9
Terminal-Bench Science 0.1 (Vals)Science8.611.4+2.9
CritPt v2 (agentic, avg@5)Science65.162.3−2.9
Vals Index v2.1Indexes54.952.1−2.8
RSI-Exam 0.1 (Full 88)Agents0.30.4+2.6
Immigration BenchLegal63.466.0+2.6
ORCA-bench Hard RCA (Fireworks 324-task subset)Agents29.532.1+2.5
Terminal-Bench 4.0Code19.717.2−2.5
CharXiv (descriptive and reasoning combined)Multimodal95.492.9−2.5
GDPval (Mercor 213-task rubric score)Professional Work83.586.0+2.5
FrontierMath Tiers 1-3 (v2) - PrivateMath68.466.0−2.5
Vibe Code Bench v1.1Code78.776.2−2.4
SpatialBench-Long V1 Clean RerunsScience28.626.2−2.4
Vals Science IndexIndexes35.838.1+2.3
Harvey HLAB — Criteria Pass RateLegal90.292.5+2.3
Clipping Bench v1.0Games596805+2.2
SkillsBench with SkillsAgents58.055.8−2.2
EpiBench V1 Clean RerunsScience21.423.6+2.2
MacroscopeBenchCode70.772.8+2.1
MedScribeHealth & Medicine84.586.5+2.0
PG-LLM-50Science0.40.3−2.0
DAYJOB: HealthcareHealth & Medicine0.82.8+2.0
LMArena Agent Arena Confirmed SuccessAgents1.3-2.6−2.0
RareDiseaseBench V1 (35 evaluations)Science2.94.8+1.9
Design Arena Overall Code CategoriesCode12991288−1.8
DTBenchGeneral QA95.797.3+1.6
CharXiv Extended (Mercor held-out)Multimodal78.977.3−1.6
Public Benefits Bench v1.1Professional Work65.366.8+1.6
GPQA DiamondGeneral QA93.995.5+1.5
Terminal-Bench 4.0 (Artificial Analysis)Code19.721.2+1.5
APEX-Accounting 1.1Finance & Accounting58.860.3+1.5
MBCT (SecureBio)Science48.950.4+1.5
CritPt (avg@5)Science18.319.7+1.4
GDPval Extended (Mercor held-out rubric score)Professional Work60.461.8+1.4
VCT (SecureBio)Science46.948.3+1.4
SnorkelUnderwrite 2.0Finance & Accounting19.117.8−1.3
MMLU ProGeneral QA90.789.4−1.3
LiveCodeBench (Vals AI public dataset)Code89.588.2−1.3
TxBench Antibody Discovery V1 Clean RerunsScience43.344.5+1.2
MortgageTax (Vals all 1,258 documents)Finance & Accounting65.364.2−1.2
Terminal-Bench 2.1 Extended (Mercor held-out)Code44.445.5+1.1
APEX-1 (Mercor, mean score)Professional Work68.867.7−1.1
DAXBench v2Code87.286.2−1.1
FrontierCode 2.0 (Artificial Analysis)Code18.719.8+1.0
ATLAS-Finance v1.1Finance & Accounting3.74.7+1.0
Snorkel Agentic Coding 2.0Code32.633.6+1.0
SpeciEval (Pooled English Runs)Safety89.990.8+1.0
SpeciEval (Refusal-Aware Retries, Historical)Safety89.990.8+1.0
AutoResearchExam (hidden-test AUARC)Agents0.50.5+1.0
BrowseComp (Mercor 130-question subset)Search83.184.0+0.9
HPCT (SecureBio)Science62.461.6−0.8
Terminal-Bench 2.1Code87.688.4+0.7
LegalBenchLegal87.086.3−0.7
TissueMicroscopyBench V1 (49 evaluations)Science3.42.7−0.7
Epoch Capabilities Index (ECI)Indexes159158−0.6
Snorkel SWE-bench CLICode6.35.7−0.6
LMArena Agent Arena SteerabilityAgents4.33.3−0.5
BioSecBench-Refusal V1 Clean RerunsSafety47.446.9−0.5
ForecastBench Preliminary Dataset Brier IndexForecasting60.561.0+0.5
AA-Omniscience IndexIndexes29.630.5+0.5
Artificial Analysis Progress Index v1Indexes320322+0.4
SciCode (main problems)Science14.214.6+0.4
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting86.085.6−0.4
MineBench — Bradley-Terry RatingGames18061811+0.3
Prophet Arena Arena Score (Fixed Context)Forecasting85.385.0−0.3
OTIS Mock AIME 2024-2025Math98.999.2+0.3
LMArena Agent Arena Tool HallucinationAgents0.41.0+0.3
LMArena Agent Arena Steering BurdenAgents-1.3-1.8−0.3
Artificial Analysis Intelligence Index v5.0.1Indexes30.831.0+0.2
SciCode 1.0.1 (Artificial Analysis)Code56.656.5−0.1
LMArena Agent Arena Bash ReliabilityAgents9.910.0+0.1
BabaIsBench Stage 1: The LakeGames86.786.70.0
Harvey's Legal Agent Benchmark - Arbitration International Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Banking FinanceLegal20.020.00.0
Harvey's Legal Agent Benchmark - Bankruptcy RestructuringLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal20.020.00.0
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 - Healthcare Life SciencesLegal20.020.00.0
Harvey's Legal Agent Benchmark - ImmigrationLegal0.00.00.0
Harvey's Legal Agent Benchmark - InsuranceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Intellectual PropertyLegal0.00.00.0
Harvey's Legal Agent Benchmark - International Trade SanctionsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Real EstateLegal0.00.00.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal20.020.00.0
Harvey's Legal Agent Benchmark - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
ObviousBenchGeneral QA100.0100.00.0
ProgramBenchCode1.01.00.0

Gemini 3.8 FlashvsGrok 4.6normalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 69 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…