Atlas

Compare Models

DeepSeek V4.1 Flash 1017 vs Muse Wonder Full

Muse Wonder Full leads by 3.7 capability points (≈95% interval +1.2 to +6.2): a clear lead over DeepSeek V4.1 Flash 1017 · P(higher) ≈ 100%

Wins 58 of 65 shared benchmarks (7 lost)

Capability vs. relative cost
DeepSeek V4.1 Flash 1017Muse 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.

DeepSeek V4.1 Flash 1017vsMuse Wonder Fullcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins7 – 58 · 65 shared

Comparing matching setups only (2 benchmarks excluded) · suites counted once
← DeepSeek V4.1 Flash 1017Muse Wonder Full →

DeepSeek V4.1 Flash 1017vsMuse Wonder Fullwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 7 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 DeepSeek V4.1 Flash 1017.

LisanBench Path LengthGames2110732190+34.4
Humanity's Last Exam DiamondGeneral QA20.848.5+27.7
Terminal-Bench 4.0Code31.858.1+26.3
RiemannBench (November 2026 rescoring)Math38.463.2+24.8
CritPt v2 (agentic, avg@5)Science50.675.1+24.6
Terminal-Bench-Science 0.1Science12.937.1+24.3
Artificial Analysis Intelligence Index v5.0Indexes28.549.7+21.2
Artificial Analysis Engineering IndexIndexes37.655.3+17.7
Artificial Analysis Coding Agent Index v2.1Indexes36.854.5+17.6
Artificial Analysis Intelligence Index v5.0.1Indexes32.650.1+17.5
LisanBench Difficulty-WeightedGames77339337+17.2
AA-Briefcase v1.1 EloAgents14161729+17.2
Code MigrationCode44.261.0+16.8
Artificial Analysis Economics IndexIndexes38.554.1+15.6
GDP.pdf (Artificial Analysis, All-pass)Professional Work17.232.6+15.4
LMArena Agent Arena Net Improvement (Steering Burden)Agents2.09.0+13.1
Artificial Analysis Finance & Accounting IndexIndexes42.255.2+13.1
Artificial Analysis Legal IndexIndexes37.750.7+13.0
AA-Omniscience IndexIndexes4.730.1+12.7
Harvey's Legal Agent BenchmarkLegal9.221.7+12.5
Artificial Analysis Intelligence Index v4.3.2Indexes45.155.8+10.7
Artificial Analysis Healthcare & Medical IndexIndexes41.351.9+10.6
Humanity's Last Exam (Text-Only)General QA43.353.6+10.3
Excel Modeling BenchmarkFinance & Accounting61.871.9+10.1
FrontierCode 2.0 (Artificial Analysis)Code28.537.8+9.3
HLAB-Vals-1.1Legal14.223.3+9.2
CritPt (avg@5)Science24.032.9+8.9
Artificial Analysis Strategy & Ops IndexIndexes49.257.9+8.7
Finance Agent v2 - PrecedentsFinance & Accounting26.535.0+8.6
Legal Research BenchLegal45.253.4+8.2
Vals Index v2.1Indexes56.664.6+8.0
LMArena Code Arena WebDevCode16641723+7.8
LMArena Agent Arena Praise vs ComplaintAgents1.215.4+7.1
Artificial Analysis Progress Index v1Indexes324343+6.8
SciCode 1.0.1 (Artificial Analysis)Code55.962.6+6.7
Vals Legal IndexIndexes71.977.9+6.0
Excel Modeling Benchmark v1.1Finance & Accounting69.275.0+5.8
Vals Index v2.2Indexes61.366.8+5.5
Finance Agent v2 - Disclosure AnalysisFinance & Accounting66.371.6+5.3
Code Migration v1.1Code67.562.8−4.7
Finance Agent v2.1Finance & Accounting62.967.6+4.6
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine26.722.2−4.4
LMArena Agent Arena Confirmed SuccessAgents2.311.1+4.4
VulcanBench Frontier v4 (functional pass@1)Code91.395.7+4.3
Finance Agent v2 - All PassFinance & Accounting45.249.0+3.8
Vals Finance IndexIndexes60.964.6+3.7
GDPxProfessional Work5.49.0+3.6
Finance Agent v2 - Earnings AnalysisFinance & Accounting75.378.6+3.3
LMArena Agent Arena Bash ReliabilityAgents2.48.3+3.0
AutomationBench-AA 1.0.6Agents68.571.3+2.8
LMArena Agent Arena Steering BurdenAgents4.09.6+2.8
Finance Agent v2 - Market AnalysisFinance & Accounting62.764.9+2.2
DeepSWE v1.1Code77.079.0+2.0
Vibe Code Bench v1.1Code87.786.0−1.7
Finance Agent v2 - AdjustmentsFinance & Accounting58.456.7−1.7
Access to Justice BenchmarkLegal73.174.7+1.6
FrontierCode 1.1 Main (Artificial Analysis)Code43.744.9+1.2
AA-LCR v1.1Search82.783.7+1.0
Finance Agent v2 - Financial ModelingFinance & Accounting22.823.7+1.0
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting79.280.1+0.9
Finance Agent v2 - ComparablesFinance & Accounting47.546.8−0.7
Bug Hunt BenchCode40.740.0−0.7
LMArena Agent Arena Tool HallucinationAgents0.30.0−0.1
Tax Agent BenchFinance & Accounting74.975.0+0.1
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting74.174.2+0.1

DeepSeek V4.1 Flash 1017vsMuse Wonder Fullnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 7 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…