Atlas

Compare Models

GPT-6 Astra vs Muse Wonder

GPT-6 Astra leads by 3.7 capability points (≈95% interval +1.2 to +6.2): a clear lead over Muse Wonder · P(higher) ≈ 100%

Wins 120 of 222 shared benchmarks (87 lost, 15 tied)

Capability vs. relative cost
GPT-6 AstraMuse WonderClosed 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.

GPT-6 AstravsMuse Wondercapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins120 – 87 · 222 shared, 15 tied

Comparing matching setups only (86 benchmarks excluded) · suites counted once
← GPT-6 AstraMuse Wonder →

GPT-6 AstravsMuse Wonderwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 45 othersVillage LLM Atlasatlas.village.dev

Margins

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

AA-PersonalityBench Informativeness indexChat & Writing29.694.3+65.4
LisanBench Path LengthGames739327973+63.9
LisanBench Difficulty-WeightedGames31138303+55.6
VoxelBench Text RatingGames26251870−53.8
HPCT Refusal (CAIS)Safety50.097.0+47.0
EBR-bench v4 (card ban, single agent)Games76.231.4−44.8
VCT-Refusal (CAIS)Safety17.660.2+42.6
ChessBench EloGames17551046−42.2
Harvey's Legal Agent Benchmark - Litigation Dispute ResolutionLegal0.040.0+40.0
Terminal-Bench Science 0.1 (Vals)Science62.922.9−40.0
DeepResearch Bench II (Mercor 130-task set)Search50.110.8−39.3
CyberBench v1.1 OverallCybersecurity41.180.4+39.3
AA-PersonalityBench Transparency indexChat & Writing94.356.2−38.5
AA-PersonalityBench Slop indexChat & Writing16.254.3+38.5
Terminal-Bench-Science 0.1Science59.521.9−37.6
Code2Video-AA EloCode1427997−34.9
AA-PersonalityBench Conviction indexChat & Writing44.879.1+34.6
NoCoT-Bench NCRI 15.2General QA15997.7−33.7
EnigmaEvalGeneral QA69.736.1−33.6
IOI (Vals v2)Code100.066.9−33.1
Mystery Game PuzzlesGames84.051.0−33.0
OpenWorldEngineer (Mean@8)Code44.212.7−31.5
AA-PersonalityBench Expressiveness indexChat & Writing58.188.6+30.8
HoneyBench v0.2 (hack rate)Safety50.881.0−30.3
TextQuests Average Cumulative HarmSafety29.220.8+28.7
CheatBenchSafety45.073.1−28.1
WeirdML v3Science42.215.2−27.0
FrontierMath Tier 4 (v2) - PrivateMath97.670.7−26.9
TextQuests Game ProgressGames74.247.6−26.6
Vals Science IndexIndexes65.139.1−26.0
Furniture Assembly BenchmarkMultimodal80.054.2−25.8
MysteryMechanismScience53.227.5−25.7
AA-PersonalityBench Warmth indexChat & Writing75.3100+25.0
Humanity's Last Exam DiamondGeneral QA63.839.0−24.8
LMArena Agent Arena Net Improvement (Steering Burden)Agents14.51.8−24.1
DataCenterBench Tier 1Agents40.016.7−23.3
RiemannBench (November 2026 rescoring)Math81.659.2−22.4
TaxCalcBench TY25Finance & Accounting44.022.0−22.0
Machiavelli Deception (CAIS)Safety76.383.6−22.0
GBENCH One-shot Coding GScoreGames79.258.1−21.2
AA-PersonalityBench Organization indexChat & Writing69.590.5+21.2
Eval Connections One-Shot with TrapsGames94.073.0−21.0
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 - Capital MarketsLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal0.020.0+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 - Intellectual PropertyLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Real EstateLegal0.020.0+20.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine41.722.2−19.4
EnterpriseBench: CoreCraftProfessional Work42.661.5+18.9
BioTIER RefusalSafety79.998.7+18.8
VIALSMultimodal49.931.5−18.4
CrossSearch (pass^3)Search60.042.0−18.0
SnorkelManufacturingProfessional Work10.027.8+17.8
Visa Agentic Commerce BenchmarkAgents64.482.2+17.8
SnorkelLegalLegal20.938.3+17.4
Decision-Bench v1.1Professional Work51.534.6−16.9
MMMU-Pro Extended (Mercor held-out)Multimodal60.043.2−16.8
LLM Creative Story WritingChat & Writing3.42.0−16.8
LMArena Agent Arena Praise vs ComplaintAgents31.7-1.8−16.7
HLAB-Vals-1.1Legal7.123.8+16.7
GameBench 2Games94.377.8−16.5
ChessBench AccuracyGames85.070.2−15.9
Terminal-Bench 4.0 (Artificial Analysis)Code54.038.4−15.7
NoCoT-Bench NCKI (kspine v3)General QA131109−15.0
ProofBench v1.1Math99.084.0−15.0
sudo L7 (October 2026 rescoring)Code31.116.1−15.0
Build-A-Chip-Bench (20-task index run)Agents15.00.3−14.7
BioSecBench-Refusal V1 Clean RerunsSafety40.855.3+14.5
FrontierCode 2.0 (Artificial Analysis)Code42.427.9−14.5
HiL-Bench (Human-in-Loop Benchmark)Agents65.379.7+14.4
HardGeoBench Avg DistanceMultimodal9081369−14.2
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal9.223.0+13.8
MacroscopeBenchCode78.064.2−13.8
MASKSafety98.584.9−13.6
AA-PersonalityBench Imagination indexChat & Writing77.290.5+13.5
Artificial Analysis Intelligence Index v5.0.1Indexes56.943.6−13.3
CritPt v2 (agentic, avg@5)Science83.770.9−12.9
DAXBench v2Code96.884.0−12.8
HardGeoBench <=100miMultimodal51.338.8−12.5
Artificial Analysis Engineering IndexIndexes61.849.6−12.2
BioMysteryBench (Vals Terminus 2)Science79.367.0−12.2
BullshitBench v2 LeaderboardSafety69.057.0−12.0
Artificial Analysis Intelligence Index v5.0Indexes55.343.3−11.9
DistressBenchSafety74.385.9+11.6
FrontierMath Tiers 1-3 (v2) - PrivateMath93.782.1−11.6
Professional Reasoning Benchmark - FinanceFinance & Accounting47.558.7+11.1
Legal Research BenchLegal39.450.5+11.1
Finance Agent v2 - ComparablesFinance & Accounting37.847.9+10.2
Terminal-Bench 4.0Code59.649.5−10.1
Professional Reasoning Benchmark - LegalLegal48.458.4+10.0
HardGeoBench <=10miMultimodal37.527.5−10.0
Humanity's Sixth SenseMultimodal53.643.7−9.9
SpatialViz-BenchMultimodal93.183.5−9.6
AA-Omniscience IndexIndexes43.725.5−9.1
WellnessBenchSafety64.055.0−9.0
Artificial Analysis Legal IndexIndexes55.646.9−8.7
Creative Writing v3 EloChat & Writing21772007−8.6
Conceptual Reasoning IndexIndexes73.967.7−8.5
Code Migration v1.1Code80.672.4−8.2
GDPval-AA v2.1Professional Work15871770+8.1
LiveBench 2.0Agents86.178.1−8.0
Tax Agent BenchFinance & Accounting63.371.2+7.8
Artificial Analysis Economics IndexIndexes57.249.5−7.7
LMCAGeneral QA64.456.8−7.5
HardGeoBench Country AccuracyMultimodal60.052.5−7.5
Finance Agent v2 - Earnings AnalysisFinance & Accounting69.576.8+7.3
IntPhys 2Multimodal77.470.2−7.2
BuseyBench SVGMultimodal7.36.6−7.2
SnorkelRevOpsProfessional Work10.918.0+7.1
ERQAMultimodal82.375.3−7.0
BioResearchBenchScience4.011.0+7.0
Vals Legal IndexIndexes68.475.3+6.9
LMArena Text Arena EloChat & Writing14741511+6.4
DAYJOB: HealthcareHealth & Medicine11.618.0+6.4
Financial Coaching BenchFinance & Accounting67.573.8+6.2
Artificial Analysis Finance & Accounting IndexIndexes57.451.1−6.2
HackerRank-BenchCode43.549.7+6.2
LMArena Agent Arena Steering BurdenAgents14.82.6−6.1
APEX-SWECode48.654.6+6.0
Humanity's Last ExamGeneral QA53.647.6−6.0
GDPval Extended (Mercor held-out rubric score)Professional Work64.170.0+5.9
APEX-1 (Mercor, mean score)Professional Work69.375.1+5.8
Epoch Capabilities Index (ECI)Indexes169162−5.8
Finance Agent v2 - All PassFinance & Accounting40.345.9+5.7
Vals Finance IndexIndexes55.661.2+5.6
Finance Agent v2 - Disclosure AnalysisFinance & Accounting61.767.3+5.6
ACCoRDGeneral QA81.776.2−5.5
CharXiv Extended (Mercor held-out)Multimodal85.179.6−5.5
Harvey HLAB — Criteria Pass RateLegal87.592.7+5.2
τ³-Banking (Mercor run, bm25_grep retrieval)Agents46.751.9+5.2
Code MigrationCode67.762.6−5.2
Finance Agent v2.1Finance & Accounting60.465.5+5.1
Artificial Analysis Healthcare & Medical IndexIndexes53.748.7−5.0
AA-Briefcase v1.1 EloAgents15691659+4.9
Artificial Analysis Progress Index v1Indexes350336−4.8
DeepSWE v1.1Code74.178.8+4.7
Excel Modeling BenchmarkFinance & Accounting71.767.1−4.6
SWE Atlas v1.1 - OverallCode69.464.8−4.6
DAYJOB: Data ScienceProfessional Work21.316.7−4.6
JobBench - release-reported mean task rewardProfessional Work66.170.5+4.4
Finance Agent v2 - AdjustmentsFinance & Accounting48.352.7+4.4
SpatialBench V1 Clean RerunsScience67.571.9+4.3
LMArena Agent Arena Confirmed SuccessAgents14.05.4−4.3
PCT Political Inconsistency (CAIS)Safety37.433.3+4.1
Snorkel Agentic Coding 2.0Code47.643.6−4.0
SciCode 1.0.1 (Artificial Analysis)Code56.560.3+3.8
SciCode (main problems)Science19.615.8−3.8
Artificial Analysis Strategy & Ops IndexIndexes58.054.5−3.5
Finance Agent v2 - Market AnalysisFinance & Accounting65.562.1−3.4
CritPt (avg@5)Science31.728.3−3.4
GDP.pdfProfessional Work34.230.8−3.4
SnorkelFinance 2.0Finance & Accounting11.615.0+3.4
Snorkel Digital Work IndexIndexes23.026.4+3.4
OSWorld 2.1 full - reported partial reward, step limit unspecifiedAgents75.872.6−3.2
ChessBench CoherenceGames97.0100.0+3.0
Snorkel WorkplaceAgentsProfessional Work14.417.4+3.0
ComplexConstraints (November 2026 rescoring)Chat & Writing70.867.9−2.9
GDPxProfessional Work9.66.8−2.8
MKA (Artificial Analysis)Professional Work50.247.5−2.8
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine21.324.0+2.7
AutomationBench-AA - release-reported strict pass rateAgents41.639.1−2.5
Dual-Arm Sim (task completion)Agents2.50.0−2.5
ProgramBench - almost solved rollout percentageCode52.550.0−2.5
APEX-Accounting 1.1Finance & Accounting64.767.1+2.4
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting68.470.8+2.4
SWE-bench Multimodal (Vals v1, 480 Tasks)Code43.345.6+2.3
DTBenchGeneral QA97.395.2−2.1
DualEntry Accounting AI BenchmarkFinance & Accounting81.283.2+2.0
Immigration BenchLegal67.769.6+1.9
APEX-Agents 1.1Professional Work75.177.0+1.9
MMLU ProGeneral QA91.589.6−1.9
HANDBOOK.md Agents v1.1Professional Work31.429.5−1.9
Finance Agent v2 - PrecedentsFinance & Accounting30.332.1+1.8
SnorkelUnderwrite 2.0Finance & Accounting21.223.0+1.8
MindCube TinyMultimodal92.791.1−1.6
CharXiv (descriptive and reasoning combined)Multimodal96.194.5−1.6
MMMU ProMultimodal86.484.9−1.5
LMArena Agent Arena Bash ReliabilityAgents7.84.8−1.5
Creative Writing v3 Rubric ScoreChat & Writing15.916.2+1.4
Toloka Arena (v3)Agents67.065.7−1.3
DAYJOB: TaxFinance & Accounting4.86.0+1.2
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting78.877.6−1.2
micro1 PharmaBenchHealth & Medicine16.117.2+1.1
Vibe Code Bench 1–100Code27.626.6−1.1
Snorkel SWE-bench CLICode12.411.4−1.0
SpeciEval (Pooled English Runs)Safety96.597.5+1.0
Finance Agent v2 - Financial ModelingFinance & Accounting21.622.5+0.9
DAYJOB: AccountingFinance & Accounting30.030.8+0.8
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting86.285.4−0.8
LMArena Agent Arena Tool HallucinationAgents1.0-0.5−0.8
Design Arena Overall Code CategoriesCode13891385−0.7
OTIS Mock AIME 2024-2025Math100.099.4−0.6
FrontierCode 1.1 Main (Artificial Analysis)Code47.046.5−0.5
Prophet Arena Arena Score (Fixed Context)Forecasting84.785.2+0.5
AutomationBench-AA 1.0.6Agents68.568.1−0.4
MissionReportProfessional Work61.862.1+0.3
DAYJOB: LawLegal36.936.6−0.3
GDPval (Mercor 213-task rubric score)Professional Work86.787.0+0.3
Artificial Analysis Intelligence Index v4.3.2Indexes52.752.4−0.3
DAYJOB: FinanceFinance & Accounting21.521.3−0.2
Excel Modeling Benchmark v1.1Finance & Accounting72.772.5−0.2
Ramp Accounting BenchFinance & Accounting48.748.70.0
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 - Corporate M&ALegal20.020.00.0
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 - International Trade SanctionsLegal0.00.00.0
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 - Trusts Estates Private ClientLegal20.020.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
OSWorld 2.1 full - reported binary completion, step limit unspecifiedAgents34.334.30.0
VulnPR-100Code46.046.00.0

GPT-6 AstravsMuse Wondernormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Mercor and 45 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…