Atlas

Compare Models

GPT-5.6 Sol vs Muse Wonder

Muse Wonder leads by 0.7 capability points (≈95% interval −1.6 to +3.0): statistically tied · P(higher) ≈ 73%

Wins 74 of 121 shared benchmarks (47 lost)

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

Benchmark wins47 – 74 · 121 shared

Comparing matching setups only (62 benchmarks excluded) · suites counted once
← GPT-5.6 SolMuse Wonder →

GPT-5.6 SolvsMuse Wonderwins by categoryData Nov 15, 2026Data: Vals AI, Center for AI Safety, Artificial Analysis and 34 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-5.6 Sol.

LisanBench Difficulty-WeightedGames14038303+73.9
HPCT Refusal (CAIS)Safety24.097.0+73.0
LisanBench Path LengthGames469927973+72.3
VCT-Refusal (CAIS)Safety11.360.2+48.9
HiL-Bench (Human-in-Loop Benchmark)Agents32.379.7+47.3
DeepResearch Bench II (Mercor 130-task set)Search51.510.8−40.7
TextQuests Average Cumulative HarmSafety31.420.8+36.1
VoxelBench Text RatingGames22971870−30.4
BioTIER RefusalSafety71.998.7+26.9
IOI (Vals v2)Code91.266.9−24.2
RiemannBench (November 2026 rescoring)Math83.259.2−24.0
Terminal-Bench 4.0 (Artificial Analysis)Code20.738.4+17.7
DAYJOB: HealthcareHealth & Medicine1.618.0+16.4
DAYJOB: FinanceFinance & Accounting5.521.3+15.8
EnterpriseBench: CoreCraftProfessional Work46.261.5+15.3
Harvey's Legal Agent BenchmarkLegal2.517.1+14.6
MindCube TinyMultimodal76.891.1+14.2
Humanity's Sixth SenseMultimodal30.043.7+13.7
EBR-bench v4 (card ban, single agent)Games44.831.4−13.3
DistressBenchSafety73.685.9+12.3
FrontierMath Tier 4 (v2) - PrivateMath82.970.7−12.2
Terminal-Bench 4.0Code37.949.5+11.6
HackerRank-BenchCode38.449.7+11.3
ChessBench CoherenceGames89.1100.0+10.9
MASKSafety95.484.9−10.5
ChessBench EloGames12181046−10.2
Eval Connections One-Shot with TrapsGames83.073.0−10.0
AA-Briefcase v1.1 EloAgents14791659+9.9
Code MigrationCode52.962.6+9.6
GBENCH One-shot Coding GScoreGames67.758.1−9.6
BullshitBench v2 LeaderboardSafety48.057.0+9.0
CursorBench 4.0Code41.750.4+8.7
Humanity's Last Exam DiamondGeneral QA30.439.0+8.6
PCT Political Inconsistency (CAIS)Safety41.733.3+8.4
DeepSWE v1.1Code70.778.8+8.1
AutomationBench-AA 1.0.6Agents60.168.1+8.0
TaxCalcBench TY25Finance & Accounting30.022.0−8.0
Design Arena Overall Code CategoriesCode13381385+7.8
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine30.022.2−7.8
NoCoT-Bench NCKI (kspine v3)General QA121109−7.8
FrontierMath Tiers 1-3 (v2) - PrivateMath89.182.1−7.0
Mystery Game PuzzlesGames58.051.0−7.0
GDPval-AA v2.1Professional Work16191770+6.7
ChessBench AccuracyGames76.470.2−6.6
APEX-SWECode48.054.6+6.6
NoCoT-Bench NCRI 15.2General QA10997.7−6.4
IntPhys 2Multimodal63.970.2+6.3
CritPt v2 (agentic, avg@5)Science77.171.1−6.0
Finance Agent v2 - PrecedentsFinance & Accounting26.232.1+5.9
MysteryMechanismScience33.327.5−5.9
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting64.970.8+5.8
LLM Creative Story WritingChat & Writing2.52.0−5.8
Artificial Analysis Intelligence Index v4.3.2Indexes47.052.4+5.4
LMArena Agent Arena Net Improvement (Steering Burden)Agents4.61.8−5.4
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine18.724.0+5.3
Excel Modeling BenchmarkFinance & Accounting72.367.1−5.3
HardGeoBench Country AccuracyMultimodal47.552.5+5.0
VIALSMultimodal26.531.5+5.0
BuseyBench SVGMultimodal6.16.6+4.9
Toloka Arena (v3)Agents61.065.7+4.7
Finance Agent v2 - All PassFinance & Accounting41.445.9+4.6
LMArena Text Arena EloChat & Writing14851511+4.5
TextQuests Game ProgressGames52.147.6−4.4
ERQAMultimodal71.275.3+4.1
LMArena Agent Arena Praise vs ComplaintAgents6.3-1.8−4.1
BioMysteryBench (Vals Terminus 2)Science71.167.0−4.1
CyberBench v1.1 OverallCybersecurity76.380.4+4.0
CrossSearch (pass^3)Search38.042.0+4.0
Finance Agent v2 - Disclosure AnalysisFinance & Accounting63.367.3+3.9
SpatialBench V1 Clean RerunsScience68.171.9+3.8
Public Benefits Bench v1.1Professional Work66.570.2+3.7
Finance Agent v2 - AdjustmentsFinance & Accounting49.252.7+3.4
Terminal-Bench-Science 0.1Science22.425.7+3.3
LMArena Agent Arena Bash ReliabilityAgents11.44.8−3.3
Tax Agent BenchFinance & Accounting68.071.2+3.2
Vals Index v2.1Indexes58.061.1+3.1
Terminal-Bench Science 0.1 (Vals)Science20.022.9+2.9
GDP.pdf (Artificial Analysis, All-pass)Professional Work27.830.6+2.8
SpatialViz-BenchMultimodal80.783.5+2.8
Finance Agent v2 - ComparablesFinance & Accounting45.347.9+2.6
Machiavelli Deception (CAIS)Safety82.783.6−2.6
Vibe Code Bench v1.1Code80.583.1+2.6
SciCode 1.0.1 (Artificial Analysis)Code57.860.3+2.5
Furniture Assembly BenchmarkMultimodal56.754.2−2.5
HardGeoBench <=100miMultimodal41.338.8−2.5
HardGeoBench <=10miMultimodal30.027.5−2.5
Finance Agent v2 - Earnings AnalysisFinance & Accounting74.476.8+2.4
Legal Research BenchLegal48.150.5+2.4
Vals Science IndexIndexes41.539.1−2.4
MMMU ProMultimodal82.784.9+2.2
GameBench 2Games80.177.8−2.2
Humanity's Last ExamGeneral QA45.547.6+2.1
Finance Agent v2 - Market AnalysisFinance & Accounting64.262.1−2.0
WellnessBenchSafety57.055.0−2.0
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting75.677.6+2.0
MacroscopeBenchCode66.164.2−1.9
HardGeoBench Avg DistanceMultimodal14291369+1.9
LiveBench 2.0Agents76.378.1+1.8
Finance Agent v2 - Financial ModelingFinance & Accounting20.722.5+1.8
CheatBenchSafety74.973.1+1.8
AA-Omniscience IndexIndexes22.025.5+1.8
Creative Writing v3 EloChat & Writing19742007+1.7
LMCAGeneral QA58.556.8−1.6
LMArena Agent Arena Confirmed SuccessAgents2.35.4+1.5
SpeciEval (Pooled English Runs)Safety99.097.5−1.5
Artificial Analysis Progress Index v1Indexes331336+1.5
LMArena Agent Arena Steering BurdenAgents5.52.6−1.5
Conceptual Reasoning IndexIndexes68.767.7−1.4
Vending-Bench 2Games96199432−1.1
Financial Coaching BenchFinance & Accounting74.973.8−1.1
ProofBench v1.1Math83.084.0+1.0
Epoch Capabilities Index (ECI)Indexes163162−0.8
LMArena Agent Arena Tool HallucinationAgents1.0-0.5−0.8
OTIS Mock AIME 2024-2025Math100.099.4−0.6
DTBenchGeneral QA95.795.2−0.5
SciCode (main problems)Science16.215.8−0.4
Prophet Arena Arena Score (Fixed Context)Forecasting84.885.2+0.3
CritPt (avg@5)Science28.628.3−0.3
ACCoRDGeneral QA75.976.2+0.3
Creative Writing v3 Rubric ScoreChat & Writing16.216.2+0.2
WeirdML v3Science15.015.2+0.2

GPT-5.6 SolvsMuse Wondernormalised score marginsData Nov 15, 2026Data: Vals AI, Center for AI Safety, Artificial Analysis and 34 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…