Atlas

Compare Models

DeepSeek V4.1 Flash vs Muse Wonder

Muse Wonder leads by 4.2 capability points (≈95% interval +2.1 to +6.3): a clear lead over DeepSeek V4.1 Flash · P(higher) ≈ 100%

Wins 108 of 131 shared benchmarks (21 lost, 2 tied)

Capability vs. relative cost
DeepSeek V4.1 FlashMuse 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.

DeepSeek V4.1 FlashvsMuse Wondercapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins21 – 108 · 131 shared, 2 tied

Comparing matching setups only (9 benchmarks excluded) · suites counted once
← DeepSeek V4.1 FlashMuse Wonder →

DeepSeek V4.1 FlashvsMuse Wonderwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 31 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Wonder minus DeepSeek V4.1 Flash.

AA-PersonalityBench Imagination indexChat & Writing31.590.5+59.6
AA-PersonalityBench Warmth indexChat & Writing41.0100+59.6
BioSecBench-Refusal V1 Clean RerunsSafety11.355.3+44.0
FrontierMath Tier 4 (v2) - PrivateMath26.870.7+43.9
DeepResearch Bench II (Mercor 130-task set)Search53.210.8−42.4
ProofBench v1.1Math54.084.0+30.0
Terminal-Bench 4.0Code19.749.5+29.8
LLM Creative Story WritingChat & Writing-0.52.0+27.7
AA-PersonalityBench Expressiveness indexChat & Writing61.988.6+26.9
IOI (Vals v2)Code40.366.9+26.7
APEX-Agents 1.1Professional Work50.577.0+26.5
BuseyBench SVGMultimodal4.16.6+25.1
Creative Writing v3 EloChat & Writing15402007+23.6
Terminal-Bench Science 0.1 (Vals)Science0.022.9+22.9
EnterpriseBench: CoreCraftProfessional Work41.561.5+20.0
Furniture Assembly BenchmarkMultimodal34.254.2+20.0
Humanity's Last Exam DiamondGeneral QA19.139.0+19.9
VIALSMultimodal12.631.5+18.9
MMMU-Pro Extended (Mercor held-out)Multimodal25.043.2+18.2
TaxCalcBench TY25Finance & Accounting4.022.0+18.0
GDP.pdf (Artificial Analysis, All-pass)Professional Work12.830.6+17.8
HieroglyphBenchMultimodal12.329.7+17.4
AA-PersonalityBench Conviction indexChat & Writing96.279.1−17.3
AA-PersonalityBench Slop indexChat & Writing37.254.3+17.3
APEX-Accounting 1.1Finance & Accounting50.167.1+17.0
Code MigrationCode45.662.6+16.9
Terminal-Bench-Science 0.1Science9.025.7+16.7
AA-Omniscience IndexIndexes-5.325.5+15.4
AA-PersonalityBench Informativeness indexChat & Writing79.194.3+15.4
FrontierMath Tiers 1-3 (v2) - PrivateMath67.482.1+14.7
τ³-Banking (Mercor run, bm25_grep retrieval)Agents37.851.9+14.1
CritPt (avg@5)Science14.328.3+14.0
Code Migration v1.1Code58.972.4+13.4
AA-Briefcase v1.1 EloAgents14151659+13.4
Artificial Analysis Intelligence Index v4.3.2Indexes39.552.4+12.9
Conceptual Reasoning IndexIndexes58.467.7+12.8
Terminal-Bench 4.0 (Artificial Analysis)Code26.838.4+11.6
Creative Writing v3 Rubric ScoreChat & Writing14.116.2+10.8
Humanity's Last ExamGeneral QA36.847.6+10.8
RuneBenchGames5.75.0−10.7
CharXiv Extended (Mercor held-out)Multimodal68.979.6+10.7
Harvey's Legal Agent BenchmarkLegal6.717.1+10.4
GBENCH One-shot Coding GScoreGames48.058.1+10.1
Excel Modeling BenchmarkFinance & Accounting57.267.1+9.9
LMCAGeneral QA47.056.8+9.9
AA-PersonalityBench Organization indexChat & Writing10090.5−9.6
APEX-1 (Mercor, mean score)Professional Work65.575.1+9.6
Bug Hunt BenchCode20.630.2+9.5
Vals Science IndexIndexes29.639.1+9.5
GameBench 2Games68.477.8+9.4
WeirdML v3Science5.815.2+9.3
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine14.724.0+9.3
Decision-Bench v1.1Professional Work25.334.6+9.3
Legal Research BenchLegal41.350.5+9.1
Tax Agent BenchFinance & Accounting62.571.2+8.7
SpatialBench V1 Clean RerunsScience63.271.9+8.7
DAXBench v2Code75.584.0+8.5
SciCode 1.0.1 (Artificial Analysis)Code51.960.3+8.4
Harvey LAB-AA v1.1 (hallucination-gated)Legal1.710.0+8.3
ACCoRDGeneral QA68.176.2+8.1
Mystery Game PuzzlesGames43.051.0+8.0
Ramp Accounting BenchFinance & Accounting40.848.7+7.9
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting69.877.6+7.8
GDPval-AA v2.1Professional Work16001770+7.5
CyberBench v1.1 OverallCybersecurity73.780.4+6.7
MacroscopeBenchCode70.764.2−6.5
MysteryMechanismScience21.227.5+6.3
LMArena Text Arena EloChat & Writing14751511+6.3
DistressBenchSafety79.885.9+6.1
Public Benefits Bench v1.1Professional Work64.370.2+6.0
Finance Agent v2 - Disclosure AnalysisFinance & Accounting61.367.3+5.9
Artificial Analysis Progress Index v1Indexes320336+5.5
DTBenchGeneral QA89.995.2+5.3
Finance Agent v2 - All PassFinance & Accounting41.145.9+4.9
STEMBench120 v1.0.0 (closed-book)Science2.52.7+4.8
Excel Modeling Benchmark v1.1Finance & Accounting67.872.5+4.7
SciCode (main problems)Science11.215.8+4.6
DeepSWE v1.1Code74.278.8+4.6
LMArena Agent Arena Chat Bash ReliabilityAgents-1.08.1+4.6
Code2Video-AA EloCode1053997−4.6
Epoch Capabilities Index (ECI)Indexes156162+4.5
CharXiv (descriptive and reasoning combined)Multimodal90.194.5+4.4
Vals Finance IndexIndexes57.161.2+4.1
Immigration BenchLegal65.669.6+4.0
Finance Agent v2 - Earnings AnalysisFinance & Accounting72.976.8+3.9
Finance Agent v2 - PrecedentsFinance & Accounting28.232.1+3.9
AA-PersonalityBench Transparency indexChat & Writing60.056.2−3.8
LMArena Code Arena WebDevCode16211649+3.7
AA-LCR Extended (Mercor held-out)Search70.874.4+3.6
Finance Agent v2 - AdjustmentsFinance & Accounting49.252.7+3.4
LMArena Agent Arena Chat Praise vs ComplaintAgents3.1-3.8−3.4
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting67.470.8+3.4
LMArena Agent Arena Work Net Improvement (Steering Burden)Agents2.53.9+3.2
LMArena Agent Arena Work Bash ReliabilityAgents0.15.8+2.8
LMArena Agent Arena Code Bash ReliabilityAgents-0.25.4+2.8
MineBench — Bradley-Terry RatingGames18221868+2.8
APEX-SWECode51.954.6+2.7
FrontierCode 1.1 Main (Artificial Analysis)Code44.046.5+2.4
Finance Agent v2 - ComparablesFinance & Accounting46.147.9+1.9
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine23.922.2−1.7
Vibe Code Bench v1.1Code84.783.1−1.7
LMArena Agent Arena Work Confirmed SuccessAgents5.78.8+1.6
LMArena Agent Arena Work Steering BurdenAgents1.54.4+1.4
MMLU ProGeneral QA88.289.6+1.4
LMArena Agent Arena Work Praise vs ComplaintAgents3.70.9−1.4
LMArena Agent Arena Chat Steering BurdenAgents-0.52.1+1.3
Finance Agent v2 - Market AnalysisFinance & Accounting63.462.1−1.2
OTIS Mock AIME 2024-2025Math98.399.4+1.1
LMArena Agent Arena Code Praise vs ComplaintAgents3.00.8−1.1
DualEntry Accounting AI BenchmarkFinance & Accounting84.283.2−1.0
Financial Coaching BenchFinance & Accounting72.873.8+1.0
LMArena Agent Arena Code Steering BurdenAgents2.34.1+0.9
SWE-bench Multimodal (Vals v1, 480 Tasks)Code46.545.6−0.8
AutomationBench-AA 1.0.6Agents68.968.1−0.8
LMArena Agent Arena Chat Net Improvement (Steering Burden)Agents1.61.9+0.8
BioMysteryBench (Vals Terminus 2)Science67.867.0−0.7
LMArena Agent Arena Code Net Improvement (Steering Burden)Agents2.73.1+0.7
Finance Agent v2 - Financial ModelingFinance & Accounting23.122.5−0.5
LMArena Agent Arena Chat Confirmed SuccessAgents4.95.9+0.5
OpenWorldEngineer (Mean@8)Code12.412.7+0.4
GDPval (Mercor 213-task rubric score)Professional Work86.787.0+0.3
LMArena Agent Arena Code Tool HallucinationAgents0.5-0.0−0.3
LMArena Agent Arena Code Confirmed SuccessAgents6.66.2−0.2
SpeciEval (Pooled English Runs)Safety97.497.5+0.1
Finance Agent v2.1Finance & Accounting65.465.5+0.1
Access to Justice BenchmarkLegal72.973.0+0.1
LMArena Agent Arena Work Tool HallucinationAgents0.50.3−0.1
LMArena Agent Arena Chat Tool HallucinationAgents0.30.5+0.1
LiveBench 2.0Agents78.078.1+0.1
BullshitBench v2 LeaderboardSafety57.057.00.0
Eval Connections One-Shot with TrapsGames73.073.00.0

DeepSeek V4.1 FlashvsMuse Wondernormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Arena and 31 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…