Atlas

Compare Models

GPT-5.6 Sol vs Muse Wonder Full

Muse Wonder Full leads by 1.8 capability points (≈95% interval −0.5 to +4.2): statistically tied · P(higher) ≈ 93%

Wins 48 of 61 shared benchmarks (13 lost)

Capability vs. relative cost
GPT-5.6 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-5.6 SolvsMuse Wonder Fullcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins13 – 48 · 61 shared

Comparing matching setups only (24 benchmarks excluded) · suites counted once
← GPT-5.6 SolMuse Wonder Full →

GPT-5.6 SolvsMuse Wonder Fullwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 8 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-5.6 Sol.

LisanBench Path LengthGames469932190+85.4
LisanBench Difficulty-WeightedGames14039337+85.0
Terminal-Bench 4.0Code37.958.1+20.2
RiemannBench (November 2026 rescoring)Math83.263.2−20.0
ExploitBench - release-reported exploit progressCybersecurity78.558.9−19.6
Harvey's Legal Agent BenchmarkLegal2.521.7+19.2
SRE Bench — Fully SolvedCode30.511.8−18.7
Humanity's Last Exam DiamondGeneral QA30.448.5+18.1
Terminal-Bench Science 0.1 (Vals)Science20.037.1+17.1
Vending-Bench 2Games961912292+16.2
Terminal-Bench-Science 0.1Science22.437.1+14.8
AA-Briefcase v1.1 EloAgents14791729+13.7
IOI (Vals v2)Code91.279.8−11.3
AutomationBench-AA 1.0.6Agents60.171.3+11.2
ProofBench v1.1Math83.094.0+11.0
GDPval-AA v2.1Professional Work16191857+10.5
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting64.974.2+9.2
Finance Agent v2 - PrecedentsFinance & Accounting26.235.0+8.8
Artificial Analysis Intelligence Index v4.3.2Indexes47.055.8+8.8
DeepSWE v1.1Code70.779.0+8.3
Finance Agent v2 - Disclosure AnalysisFinance & Accounting63.371.6+8.2
LMArena Agent Arena Net Improvement (Steering Burden)Agents4.69.0+8.2
Code MigrationCode52.961.0+8.1
MASKSafety95.487.3−8.1
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine30.022.2−7.8
Finance Agent v2 - All PassFinance & Accounting41.449.0+7.7
Finance Agent v2 - AdjustmentsFinance & Accounting49.256.7+7.4
Public Benefits Bench v1.1Professional Work66.573.7+7.2
Vals Science IndexIndexes41.548.5+7.1
Tax Agent BenchFinance & Accounting68.075.0+7.0
Vals Index v2.1Indexes58.064.6+6.6
Humanity's Last Exam (Text-Only)General QA47.353.6+6.3
MedCodeHealth & Medicine44.049.9+6.0
Vibe Code Bench v1.1Code80.586.0+5.5
Legal Research BenchLegal48.153.4+5.3
SciCode 1.0.1 (Artificial Analysis)Code57.862.6+4.9
GDP.pdf (Artificial Analysis, All-pass)Professional Work27.832.6+4.8
LMArena Agent Arena Praise vs ComplaintAgents6.315.4+4.6
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting75.680.1+4.4
LMArena Agent Arena Confirmed SuccessAgents2.311.1+4.4
CritPt (avg@5)Science28.632.9+4.3
Finance Agent v2 - Earnings AnalysisFinance & Accounting74.478.6+4.2
Artificial Analysis Progress Index v1Indexes331343+4.1
MedScribeHealth & Medicine85.289.3+4.1
AA-Omniscience IndexIndexes22.030.1+4.1
CyberGymCybersecurity84.588.4+3.9
Finance Agent v2 - Financial ModelingFinance & Accounting20.723.7+3.0
CyberBench v1.1 OverallCybersecurity76.373.6−2.7
BioMysteryBench (Vals Terminus 2)Science71.173.3+2.2
LMArena Agent Arena Steering BurdenAgents5.59.6+2.0
CritPt v2 (agentic, avg@5)Science77.175.1−2.0
MysteryMechanismScience33.335.1+1.8
LMArena Agent Arena Bash ReliabilityAgents11.48.3−1.6
Finance Agent v2 - ComparablesFinance & Accounting45.346.8+1.5
ProgramBenchCode1.50.5−1.0
Finance Agent v2 - Market AnalysisFinance & Accounting64.264.9+0.7
LMArena Agent Arena Tool HallucinationAgents1.00.0−0.5
LegalBenchLegal87.087.4+0.5
Excel Modeling BenchmarkFinance & Accounting72.371.9−0.4
AA-LCR v1.1Search84.083.7−0.3
Financial Coaching BenchFinance & Accounting74.975.0+0.1

GPT-5.6 SolvsMuse Wonder Fullnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 8 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…