Atlas

Compare Models

Claude Sonnet 5.5 vs Grok 4.6

Claude Sonnet 5.5 leads by 6.6 capability points (≈95% interval +4.3 to +8.9): a clear lead over Grok 4.6 · P(higher) ≈ 100%

Wins 157 of 208 shared benchmarks (30 lost, 21 tied)

Capability vs. relative cost
Claude Sonnet 5.5Grok 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.

Claude Sonnet 5.5vsGrok 4.6capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins157 – 30 · 208 shared, 21 tied

Comparing matching setups only (4 benchmarks excluded) · suites counted once
← Claude Sonnet 5.5Grok 4.6 →

Claude Sonnet 5.5vsGrok 4.6wins by categoryData Nov 15, 2026Data: Vals AI, Mercor, 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 Grok 4.6 minus Claude Sonnet 5.5.

KernelBench Mega Normalized (RTX PRO 6000)Code90.60.0−90.6
AA-PersonalityBench Organization indexChat & Writing84.814.3−71.2
AA-PersonalityBench Imagination indexChat & Writing84.820.0−65.4
Medical Long Context Reasoning (MLCR-AA)Health & Medicine75.012.2−62.8
Terminal-Bench-Science 0.1Science52.42.9−49.5
KernelBench CUDA Normalized (RTX PRO 6000)Code87.037.7−49.3
ProofBench v1.1Math100.051.0−49.0
FrontierMath Tier 4 (v2) - PrivateMath80.531.7−48.8
AA-PersonalityBench Warmth indexChat & Writing73.331.5−42.3
ChartographyMultimodal61.620.5−41.1
AA-PersonalityBench Informativeness indexChat & Writing84.844.8−40.4
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal0.040.0+40.0
Terminal-Bench 4.0 (Artificial Analysis)Code57.117.2−39.9
FrontierSWE V2Code61.925.3−36.6
Terminal-Bench 4.0Code61.825.3−36.5
Terminal-Bench Science 0.1 (Vals)Science47.111.4−35.7
IOI (Vals v2)Code83.147.6−35.4
Furniture Assembly BenchmarkMultimodal75.040.0−35.0
HardGeoBench Avg DistanceMultimodal6831807−34.7
AA-PersonalityBench Conviction indexChat & Writing58.192.4+34.6
Humanity's Last Exam DiamondGeneral QA53.321.5−31.8
Mystery Game PuzzlesGames65.034.0−31.0
LLM Chess EloGames1195443−30.9
LMArena Agent Arena Net Improvement (Steering Burden)Agents16.0-0.2−30.5
Arena Image-to-WebDevCode17441607−29.9
Olam Social Poker EloGames15261499−29.3
AA-PersonalityBench Transparency indexChat & Writing10071.4−28.8
InferenceBenchCode9.16.7−28.8
Bio Capabilities Index (SecureBio)Indexes175137−27.6
LAB-Bench LitQA2 (SecureBio)Science55.828.9−26.9
Chess Puzzles (task version 2.0)Games59.033.0−26.0
SWE Atlas Codebase QnA Extended (Mercor held-out)Code44.719.3−25.4
Code MigrationCode69.844.6−25.3
GeneBench-Pro Public Release SubsetScience39.015.0−24.0
Artificial Analysis Intelligence Index v5.0.1Indexes54.531.0−23.5
SAGEMath51.828.9−22.9
FrontierMath Tiers 1-3 (v2) - PrivateMath88.866.0−22.8
GameBench 2Games91.468.7−22.7
Harvey HLAB — Criteria Pass RateLegal70.092.5+22.6
HealthBench ProfessionalHealth & Medicine70.848.5−22.3
ARC-AGI-3 (Semi-Private)General QA24.32.1−22.2
ALE-BenchCode24481822−22.2
Bug Hunt BenchCode48.927.3−21.6
DeepsecBench — Revalidated V1 (233 issues)Code37.516.1−21.4
Artificial Analysis Coding Agent Index v1.5Indexes68.447.0−21.4
SWE-bench Verified Extended (Mercor held-out)Code57.378.0+20.7
VIALSMultimodal41.020.5−20.5
LMArena Code Arena WebDevCode17731618−20.2
LAB-Bench ProtocolQA (SecureBio)Science88.368.1−20.2
Conceptual Reasoning IndexIndexes77.162.5−20.0
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Banking FinanceLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Litigation Dispute ResolutionLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal20.00.0−20.0
LMCAGeneral QA68.348.5−19.8
SWE-Marathon v1.1Code51.331.9−19.4
BuseyBench SVGMultimodal6.54.6−19.2
Concept Synth INDUCTION Challenge100 Round 1General QA44.025.0−19.0
Roboflow Vision Evals (October 2026, five tasks)Multimodal77.659.1−18.5
MysteryMechanismScience49.130.6−18.5
MetagenomicsBench V0Science60.042.0−18.0
MetagenomicsBench V1Science60.042.0−18.0
HardGeoBench <=100miMultimodal50.032.5−17.5
Humanity's Last ExamGeneral QA56.939.7−17.2
ARC-AGI-2General QA83.967.1−16.8
LMArena Agent Arena Praise vs ComplaintAgents30.5-2.5−16.5
HardGeoBench <=10miMultimodal36.320.0−16.2
AA-AnalystAgent (pass^5)Professional Work57.541.3−16.2
WCB (SecureBio)Science66.550.3−16.2
Vibe Code Bench v1.1Code92.476.2−16.2
ScBench-Long V1 Clean RerunsScience31.815.9−15.9
FrontierCode 2.0 (Artificial Analysis)Code35.619.8−15.9
HPCT (SecureBio)Science77.261.6−15.6
Roboflow Vision Evals (September 2026 protocol)Multimodal83.868.7−15.1
Eval Connections One-Shot with TrapsGames76.091.0+15.0
HardGeoBench Country AccuracyMultimodal58.843.8−15.0
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine32.017.3−14.7
GDP.pdfProfessional Work31.417.2−14.2
CursorBench 4.0Code55.541.4−14.1
Concept Synth INDUCTION Challenge64 Round 1General QA50.035.9−14.1
ACCoRDGeneral QA85.771.6−14.1
Veracode GenAI Code Security (Fall 2026)Cybersecurity72.058.0−14.0
WellnessBenchSafety57.043.0−14.0
APEX-1 (Mercor, mean score)Professional Work53.967.7+13.8
Excel Modeling BenchmarkFinance & Accounting75.762.7−13.0
CritPt v2 (agentic, avg@5)Science75.162.3−12.9
Vending-Bench 2Games111659047−12.8
DAYJOB: HealthcareHealth & Medicine15.22.8−12.4
SWE-bench VerifiedCode96.784.6−12.1
GBENCH One-shot Coding GScoreGames67.455.4−12.0
CritPt (avg@5)Science31.419.7−11.7
Artificial Analysis Intelligence Index v4.3.2Indexes56.044.3−11.7
MineBench — Bradley-Terry RatingGames19981811−11.3
VariantBench V1 Clean RerunsScience51.740.4−11.3
DAYJOB: FinanceFinance & Accounting24.012.8−11.2
LMArena Agent Arena Confirmed SuccessAgents19.5-2.6−11.0
AA-Briefcase v1.1 EloAgents17401540−11.0
TxBench-Antibody-Development V1 (60 evaluations)Science36.726.1−10.6
Artificial Analysis Progress Index v1Indexes351322−10.2
Design Arena Overall Code CategoriesCode13491288−10.1
AA-PersonalityBench Expressiveness indexChat & Writing73.363.8−9.6
SpatialBench-Long V1 Clean RerunsScience36.527.4−9.1
GDPval Extended (Mercor held-out rubric score)Professional Work70.961.8−9.1
ARC-AGI-1 Semi-PrivateGeneral QA96.587.5−9.0
APEX-SWECode66.457.7−8.7
AA-LCR Extended (Mercor held-out)Search73.865.2−8.6
ChessBench AccuracyGames75.867.8−8.6
MBCT (SecureBio)Science58.850.4−8.4
τ³-Banking (Mercor run, bm25_grep retrieval)Agents54.045.7−8.3
MedCodeHealth & Medicine52.944.7−8.2
RedlineBenchLegal40.348.5+8.2
LMArena Agent Arena Steering BurdenAgents14.5-1.8−8.1
Toloka Arena (v3)Agents64.056.0−8.0
AA-PersonalityBench Slop indexChat & Writing27.735.3+7.7
BioSecBench-Function V1 Clean RerunsScience51.744.0−7.7
LiveBench 2.0Agents82.574.9−7.5
MedXpertQA MM Extended (Mercor held-out)Health & Medicine55.648.4−7.2
CharXiv Extended (Mercor held-out)Multimodal84.277.3−6.9
Epoch Capabilities Index (ECI)Indexes166158−6.8
EpiBench V1 Clean RerunsScience30.223.6−6.6
APEX-Agents 1.1Professional Work83.777.3−6.4
BioSecBench-Structure V1 (58 evaluations)Science21.815.5−6.3
Blueprint-Bench 2Multimodal0.40.3−6.1
YC-BenchGames17413151568241−6.0
TxBench Oligo Discovery V1 (113 evaluations)Science51.946.0−5.9
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science17.311.5−5.7
TxBench Antibody Discovery V1 (100 evaluations)Science57.351.7−5.7
MedXpertQA MM (200-question set)Health & Medicine80.775.2−5.5
ProgramBenchCode6.51.0−5.5
SpatialBench V1 Clean RerunsScience72.867.3−5.5
MMMU ProMultimodal85.880.5−5.3
EnterpriseBench: CoreCraftProfessional Work70.865.6−5.2
AutomationBench-AA 1.0.6Agents71.867.0−4.8
RuneBenchGames5.55.8+4.8
LMArena Agent Arena Bash RecoveryAgents15.05.7−4.7
ChessBench CoherenceGames64.669.2+4.6
MedScribeHealth & Medicine91.186.5−4.6
NoCoT-Bench NCRI 15.2General QA95.186.8−4.6
SciCode 1.0.1 (Artificial Analysis)Code61.056.5−4.5
APEX-Accounting 1.1Finance & Accounting64.560.3−4.2
VCT (SecureBio)Science52.548.3−4.2
ChessBenchGames39.343.3+4.0
DeepSWE v1.1Code70.566.7−3.8
GDPval-AA v2.1Professional Work17431658−3.7
Access to Justice BenchmarkLegal74.971.3−3.6
SM BenchSafety85.489.0+3.6
CharXiv (descriptive and reasoning combined)Multimodal96.192.9−3.2
NoCoT-Bench NCKI (kspine v3)General QA107112+3.2
FrontierCode 1.1 Extended - ScoreCode64.461.3−3.1
Immigration BenchLegal69.066.0−3.0
FrontierCode 1.1 Extended - Pass RateCode69.967.0−2.8
LAB-Bench CloningScenarios (SecureBio)Science97.094.2−2.8
SimpleQA Verified (Epoch anti-abstention)General QA46.549.3+2.8
LMArena Text Arena EloChat & Writing14721456−2.8
TxBench Preclinical Pharmacology V1 Clean RerunsScience64.361.7−2.7
CVDP (517-problem set)Code60.658.0−2.6
BrowseComp (Mercor 130-question subset)Search86.684.0−2.6
AdvancedIFChat & Writing83.480.9−2.5
ChessBench EloGames857891+2.0
SWE-bench MultilingualCode90.992.7+1.8
SWE-TogetherCode62.460.6−1.8
MMLU ProGeneral QA91.089.4−1.6
LMArena Agent Arena Bash ReliabilityAgents12.710.0−1.3
AA-LCR v1.1Search79.781.0+1.3
ScBench V1 Clean RerunsScience60.359.1−1.2
micro1 PharmaBenchHealth & Medicine11.412.5+1.1
Terminal-Bench 2.1Code87.388.4+1.1
AA-Omniscience IndexIndexes32.330.5−0.9
Clipping Bench v1.0Games886805−0.9
OTIS Mock AIME 2024-2025Math100.099.2−0.8
MMLU-Bio (SecureBio)Science91.091.8+0.8
Extended NYT Connections 940 (Quadratic v1)Games80.580.0−0.5
GDPval (Mercor 213-task rubric score)Professional Work86.586.0−0.5
Public Benefits Bench v1.1Professional Work67.266.8−0.3
Prophet Arena Arena Score (Fixed Context)Forecasting85.385.0−0.3
GPQA-Bio (SecureBio)Science87.687.4−0.2
SpeciEval (Refusal-Aware Retries, Historical)Safety91.090.8−0.1
SpeciEval (Pooled English Runs)Safety91.090.8−0.1
Vision Arena OverallMultimodal12631263+0.1
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting85.785.6−0.1
ForecastBench Preliminary Dataset Brier IndexForecasting61.161.0−0.1
LMArena Agent Arena Tool HallucinationAgents0.91.0+0.1
AIME 2025Math100.0100.00.0
DAXBench v2Code86.286.20.0
DTBenchGeneral QA97.397.30.0
Harvey's Legal Agent Benchmark - Arbitration International Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Bankruptcy RestructuringLegal0.00.00.0
Harvey's Legal Agent Benchmark - Capital MarketsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal20.020.00.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal0.00.00.0
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 - 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 - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
ObviousBenchGeneral QA100.0100.00.0
SimpleBenchGeneral QA75.975.90.0
SWE Atlas - Codebase QnACode50.350.30.0
Terminal-Bench 2.1 Extended (Mercor held-out)Code45.545.50.0

Claude Sonnet 5.5vsGrok 4.6normalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, 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…