Atlas

Compare Models

Muse Wonder vs Muse Wonder Full

Muse Wonder Full leads by 1.1 capability points (≈95% interval −1.3 to +3.5): statistically tied · P(higher) ≈ 82%

Wins 80 of 87 shared benchmarks (6 lost, 1 tied)

Capability vs. relative cost
Muse WonderMuse 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.

Muse WondervsMuse Wonder Fullcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins6 – 80 · 87 shared, 1 tied

Comparing matching setups only · suites counted once
← Muse WonderMuse Wonder Full →

Muse WondervsMuse Wonder Fullwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 10 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 Muse Wonder.

Vending-Bench 2Games943212292+17.3
Terminal-Bench Science 0.1 (Vals)Science22.937.1+14.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents1.89.0+13.6
LisanBench Path LengthGames2797332190+13.1
IOI (Vals v2)Code66.979.8+12.9
Terminal-Bench-Science 0.1Science25.737.1+11.4
LisanBench Difficulty-WeightedGames83039337+11.1
Artificial Analysis Coding Agent Index v2.1Indexes44.554.5+10.0
ProofBench v1.1Math84.094.0+10.0
FrontierCode 2.0 (Artificial Analysis)Code27.937.8+9.9
Bug Hunt BenchCode30.240.0+9.8
LMArena Code Arena WebDevCode16491723+9.8
Code Migration v1.1Code72.462.8−9.6
Humanity's Last Exam DiamondGeneral QA39.048.5+9.5
Vals Science IndexIndexes39.148.5+9.4
ProgramBench - almost solved rollout percentageCode50.059.0+9.0
LMArena Agent Arena Praise vs ComplaintAgents-1.815.4+8.6
Terminal-Bench 4.0Code49.558.1+8.6
MysteryMechanismScience27.535.1+7.7
Artificial Analysis Coding Agent Index v2.0Indexes52.058.9+6.9
CyberBench v1.1 OverallCybersecurity80.473.6−6.7
Artificial Analysis Intelligence Index v5.0.1Indexes43.650.1+6.5
Artificial Analysis Intelligence Index v5.0Indexes43.349.7+6.4
AutomationBench-AA - release-reported strict pass rateAgents39.145.4+6.3
BioMysteryBench (Vals Terminus 2)Science67.073.3+6.3
OSWorld 2.1 full - reported partial reward, step limit unspecifiedAgents72.678.4+5.8
Artificial Analysis Engineering IndexIndexes49.655.3+5.7
OSWorld 2.1 full - reported binary completion, step limit unspecifiedAgents34.339.4+5.1
Excel Modeling BenchmarkFinance & Accounting67.171.9+4.8
JobBench - release-reported mean task rewardProfessional Work70.575.3+4.8
Artificial Analysis Economics IndexIndexes49.554.1+4.6
Harvey's Legal Agent BenchmarkLegal17.121.7+4.6
CritPt (avg@5)Science28.332.9+4.6
SWE-Marathon - release-reported version unspecifiedCode53.157.5+4.4
Artificial Analysis Coding Agent Index v1.5Indexes63.768.0+4.4
Finance Agent v2 - Disclosure AnalysisFinance & Accounting67.371.6+4.3
Artificial Analysis Finance & Accounting IndexIndexes51.155.2+4.1
CritPt v2 (agentic, avg@5)Science71.175.1+4.0
RiemannBench (November 2026 rescoring)Math59.263.2+4.0
Finance Agent v2 - AdjustmentsFinance & Accounting52.756.7+4.0
AA-Briefcase v1.1 EloAgents16591729+3.9
Artificial Analysis Legal IndexIndexes46.950.7+3.9
Tax Agent BenchFinance & Accounting71.275.0+3.8
GDPval-AA v2.1Professional Work17701857+3.8
Humanity's Last Exam (Text-Only)General QA50.153.6+3.5
Public Benefits Bench v1.1Professional Work70.273.7+3.5
LMArena Agent Arena Steering BurdenAgents2.69.6+3.5
Vals Index v2.1Indexes61.164.6+3.5
Artificial Analysis Strategy & Ops IndexIndexes54.557.9+3.4
Artificial Analysis Intelligence Index v4.3.2Indexes52.455.8+3.4
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting70.874.2+3.4
Vals Finance IndexIndexes61.264.6+3.3
AutomationBench-AA 1.0.6Agents68.171.3+3.2
Artificial Analysis Healthcare & Medical IndexIndexes48.751.9+3.2
Finance Agent v2 - All PassFinance & Accounting45.949.0+3.1
Finance Agent v2 - PrecedentsFinance & Accounting32.135.0+2.9
Vibe Code Bench v1.1Code83.186.0+2.9
Legal Research BenchLegal50.553.4+2.9
LMArena Agent Arena Confirmed SuccessAgents5.411.1+2.9
Finance Agent v2 - Market AnalysisFinance & Accounting62.164.9+2.8
Artificial Analysis Progress Index v1Indexes336343+2.6
Vals Legal IndexIndexes75.377.9+2.6
Excel Modeling Benchmark v1.1Finance & Accounting72.575.0+2.5
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting77.680.1+2.5
GDP.pdfProfessional Work30.833.2+2.4
MASKSafety84.987.3+2.4
AA-Omniscience IndexIndexes25.530.1+2.3
SciCode 1.0.1 (Artificial Analysis)Code60.362.6+2.3
GDPxProfessional Work6.89.0+2.2
Finance Agent v2.1Finance & Accounting65.567.6+2.0
Vals Index v2.2Indexes64.966.8+1.9
Finance Agent v2 - Earnings AnalysisFinance & Accounting76.878.6+1.8
HealthBench Professional (Length-Adjusted)Health & Medicine60.562.3+1.8
LMArena Agent Arena Bash ReliabilityAgents4.88.3+1.7
Access to Justice BenchmarkLegal73.074.7+1.7
SWE-bench Multimodal (Vals v1, 480 Tasks)Code45.647.3+1.7
Immigration BenchLegal69.671.2+1.6
FrontierCode 1.1 Main (Artificial Analysis)Code46.544.9−1.6
Code MigrationCode62.661.0−1.5
Finance Agent v2 - Financial ModelingFinance & Accounting22.523.7+1.2
Financial Coaching BenchFinance & Accounting73.875.0+1.2
Finance Agent v2 - ComparablesFinance & Accounting47.946.8−1.1
SWE-Marathon v1.1Code52.553.1+0.6
HLAB-Vals-1.1Legal23.823.3−0.4
LMArena Agent Arena Tool HallucinationAgents-0.50.0+0.3
DeepSWE v1.1Code78.879.0+0.2
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine22.222.20.0

Muse WondervsMuse Wonder Fullnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 10 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…