Atlas

Compare Models

GPT-6.1 Sol vs Muse Wonder Full

GPT-6.1 Sol leads by 1.6 capability points (≈95% interval −0.9 to +4.2): statistically tied · P(higher) ≈ 89%

Muse Wonder Full wins 43 of 77 shared benchmarks (34 lost)

Capability vs. relative cost
GPT-6.1 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-6.1 SolvsMuse Wonder Fullcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins34 – 43 · 77 shared

Comparing matching setups only (34 benchmarks excluded) · suites counted once
← GPT-6.1 SolMuse Wonder Full →

GPT-6.1 SolvsMuse Wonder Fullwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 5 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-6.1 Sol.

LisanBench Path LengthGames873632190+72.9
LisanBench Difficulty-WeightedGames40049337+57.1
SRE Bench — Fully SolvedCode50.811.8−38.9
CyberBench v1.1 OverallCybersecurity39.373.6+34.3
Terminal-Bench-Science 0.1Science58.137.1−21.0
HLAB-Vals-1.1Legal6.323.3+17.1
IOI (Vals v2)Code96.979.8−17.1
Harvey's Legal Agent BenchmarkLegal5.421.7+16.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine38.322.2−16.1
RiemannBench (November 2026 rescoring)Math79.263.2−16.0
Terminal-Bench Science 0.1 (Vals)Science52.937.1−15.7
Legal Research BenchLegal38.553.4+14.9
Code Migration v1.1Code77.362.8−14.5
Public Benefits Bench v1.1Professional Work59.373.7+14.4
Finance Agent v2 - Disclosure AnalysisFinance & Accounting58.471.6+13.1
Tax Agent BenchFinance & Accounting62.375.0+12.7
Finance Agent v2 - AdjustmentsFinance & Accounting44.756.7+12.0
MysteryMechanismScience46.435.1−11.3
Vals Science IndexIndexes59.648.5−11.1
GDPval-AA v2.1Professional Work16071857+11.0
MASKSafety98.387.3−11.0
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting63.274.2+11.0
Humanity's Last Exam DiamondGeneral QA58.848.5−10.3
AA-Briefcase v1.1 EloAgents15551729+9.6
Finance Agent v2 - PrecedentsFinance & Accounting25.835.0+9.2
Vals Legal IndexIndexes68.777.9+9.1
Finance Agent v2 - All PassFinance & Accounting39.949.0+9.1
Access to Justice BenchmarkLegal65.874.7+8.8
Vals Finance IndexIndexes55.864.6+8.7
CritPt v2 (agentic, avg@5)Science82.675.1−7.4
Finance Agent v2 - Earnings AnalysisFinance & Accounting71.578.6+7.1
SciCode 1.0.1 (Artificial Analysis)Code55.862.6+6.8
LMArena Agent Arena Net Improvement (Steering Burden)Agents12.59.0−6.6
Financial Coaching BenchFinance & Accounting68.575.0+6.4
BioMysteryBench (Vals Terminus 2)Science79.673.3−6.3
LMArena Agent Arena Praise vs ComplaintAgents27.515.4−6.0
Finance Agent v2.1Finance & Accounting61.867.6+5.8
AA-Omniscience IndexIndexes41.530.1−5.7
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting74.580.1+5.5
ProofBench v1.1Math99.094.0−5.0
Artificial Analysis Intelligence Index v5.0Indexes54.549.7−4.8
AutomationBench-AA 1.0.6Agents66.671.3+4.8
Artificial Analysis Intelligence Index v5.0.1Indexes54.650.1−4.5
FrontierCode 1.1 Main (Artificial Analysis)Code49.144.9−4.2
Finance Agent v2 - ComparablesFinance & Accounting42.646.8+4.2
Code MigrationCode65.161.0−4.1
Artificial Analysis Engineering IndexIndexes59.455.3−4.1
Artificial Analysis Intelligence Index v4.3.2Indexes51.855.8+4.0
Excel Modeling Benchmark v1.1Finance & Accounting71.175.0+3.9
SWE-bench Multimodal (Vals v1, 480 Tasks)Code43.547.3+3.8
Immigration BenchLegal67.871.2+3.4
Vals Index v2.1Indexes61.264.6+3.4
Terminal-Bench 4.0Code55.158.1+3.0
Artificial Analysis Legal IndexIndexes53.750.7−2.9
Vibe Code Bench v1.1Code88.986.0−2.9
MedScribeHealth & Medicine86.589.3+2.9
FrontierCode 2.0 (Artificial Analysis)Code40.637.8−2.9
Vals Index v2.2Indexes63.966.8+2.8
ProgramBenchCode3.00.5−2.5
Artificial Analysis Economics IndexIndexes56.354.1−2.2
LMArena Agent Arena Steering BurdenAgents13.49.6−1.9
Artificial Analysis Progress Index v1Indexes348343−1.8
LMArena Agent Arena Bash ReliabilityAgents5.08.3+1.7
Excel Modeling BenchmarkFinance & Accounting70.871.9+1.1
CritPt (avg@5)Science31.732.9+1.1
MedCodeHealth & Medicine48.849.9+1.1
Artificial Analysis Finance & Accounting IndexIndexes56.255.2−1.0
Humanity's Last Exam (Text-Only)General QA52.653.6+1.0
Finance Agent v2 - Financial ModelingFinance & Accounting23.123.7+0.6
GDP.pdf (Artificial Analysis, All-pass)Professional Work32.032.6+0.6
GDPxProfessional Work8.49.0+0.6
LMArena Agent Arena Tool HallucinationAgents1.00.0−0.5
Finance Agent v2 - Market AnalysisFinance & Accounting64.564.9+0.4
AA-LCR v1.1Search84.083.7−0.3
LMArena Agent Arena Confirmed SuccessAgents11.711.1−0.3
Artificial Analysis Strategy & Ops IndexIndexes57.657.9+0.3
Artificial Analysis Healthcare & Medical IndexIndexes52.051.9−0.1

GPT-6.1 SolvsMuse Wonder Fullnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 5 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…