Atlas

Compare Models

GPT-6 Sol vs Muse Wonder Full

Muse Wonder Full leads by 0.9 capability points (≈95% interval −1.5 to +3.3): statistically tied · P(higher) ≈ 77%

Wins 45 of 58 shared benchmarks (12 lost, 1 tied)

Capability vs. relative cost
GPT-6 SolMuse Wonder FullClosed 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 SolvsMuse Wonder Fullcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins12 – 45 · 58 shared, 1 tied

Comparing matching setups only (25 benchmarks excluded) · suites counted once
← GPT-6 SolMuse Wonder Full →

GPT-6 SolvsMuse Wonder Fullwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 4 othersVillage LLM Atlasatlas.village.dev

Margins

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

LisanBench Path LengthGames822232190+74.5
LisanBench Difficulty-WeightedGames31189337+66.6
Legal Research BenchLegal28.853.4+24.5
Tax Agent BenchFinance & Accounting53.075.0+22.0
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting53.674.2+20.6
Harvey's Legal Agent BenchmarkLegal1.721.7+20.0
SRE Bench — Fully SolvedCode30.511.8−18.7
Public Benefits Bench v1.1Professional Work56.673.7+17.1
Finance Agent v2 - Disclosure AnalysisFinance & Accounting55.671.6+15.9
GDPval-AA v2.1Professional Work15141857+15.1
AA-Briefcase v1.1 EloAgents14761729+13.9
Terminal-Bench 4.0Code44.458.1+13.6
Finance Agent v2 - AdjustmentsFinance & Accounting43.256.7+13.4
Finance Agent v2 - All PassFinance & Accounting36.049.0+13.1
Vending-Bench 2Games1442812292−12.9
Financial Coaching BenchFinance & Accounting62.675.0+12.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine33.922.2−11.7
ProofBench v1.1Math83.094.0+11.0
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting70.080.1+10.1
AutomationBench-AA 1.0.6Agents61.771.3+9.6
Harvey LAB-AA v1.1 (hallucination-gated)Legal3.613.1+9.4
Artificial Analysis Intelligence Index v4.3.2Indexes47.655.8+8.2
Humanity's Last Exam (Text-Only)General QA46.353.6+7.3
MedScribeHealth & Medicine82.089.3+7.3
Terminal-Bench Science 0.1 (Vals)Science30.037.1+7.1
Terminal-Bench-Science 0.1Science30.037.1+7.1
Vals Index v2.1Indexes57.564.6+7.0
Finance Agent v2 - Market AnalysisFinance & Accounting58.064.9+7.0
Finance Agent v2 - PrecedentsFinance & Accounting28.535.0+6.5
Finance Agent v2 - Earnings AnalysisFinance & Accounting72.478.6+6.2
Finance Agent v2 - ComparablesFinance & Accounting40.846.8+6.0
Artificial Analysis Intelligence Index v5.0.1Indexes44.250.1+5.9
CritPt v2 (agentic, avg@5)Science80.975.1−5.7
GDP.pdfProfessional Work28.033.2+5.2
SciCode 1.0.1 (Artificial Analysis)Code57.662.6+5.0
MysteryMechanismScience30.235.1+5.0
Finance Agent v2 - Financial ModelingFinance & Accounting19.323.7+4.4
CyberBench v1.1 OverallCybersecurity78.073.6−4.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents6.99.0+3.9
Code MigrationCode57.261.0+3.8
Vals Science IndexIndexes45.048.5+3.5
LMArena Agent Arena Confirmed SuccessAgents4.511.1+3.3
Humanity's Last Exam DiamondGeneral QA45.248.5+3.3
Artificial Analysis Progress Index v1Indexes334343+3.2
MedCodeHealth & Medicine47.149.9+2.9
IOI (Vals v2)Code82.679.8−2.8
LMArena Agent Arena Praise vs ComplaintAgents10.015.4+2.7
FrontierCode 2.0 (Artificial Analysis)Code35.637.8+2.2
CritPt (avg@5)Science30.932.9+2.0
Vibe Code Bench v1.1Code87.886.0−1.8
AA-Omniscience IndexIndexes27.130.1+1.5
ProgramBenchCode2.00.5−1.5
BioMysteryBench (Vals Terminus 2)Science74.873.3−1.5
LMArena Agent Arena Bash ReliabilityAgents10.98.3−1.3
LMArena Agent Arena Tool HallucinationAgents1.00.0−0.5
Excel Modeling BenchmarkFinance & Accounting71.571.9+0.4
LMArena Agent Arena Steering BurdenAgents10.19.6−0.3
AA-LCR v1.1Search83.783.70.0

GPT-6 SolvsMuse Wonder Fullnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 4 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…