Atlas

Compare Models

DeepSeek V4.1 Flash 1017 vs Muse Ember

Muse Ember leads by 1.4 capability points (≈95% interval −1.2 to +4.0): statistically tied · P(higher) ≈ 86%

Wins 36 of 52 shared benchmarks (16 lost)

Capability vs. relative cost
DeepSeek V4.1 Flash 1017Muse EmberClosed weightsOpen weights

Benchmark wins16 – 36 · 52 shared

Comparing matching setups only (5 benchmarks excluded) · suites counted once
← DeepSeek V4.1 Flash 1017Muse Ember →
Data: Artificial Analysis, Vals AI, Arena and 6 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Ember minus DeepSeek V4.1 Flash 1017.

Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering68.547.1−21.4
RiemannBench (November 2026 rescoring)Math38.453.6+15.2
DAXBench v2Software › Data & ML77.762.8−14.9
CritPt v2 (agentic, avg@5)Science › Physical sciences50.662.3+11.7
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax12.524.2+11.7
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations41.653.3+11.7
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine6.017.2+11.2
HLAB-Vals-1.1Professional › Legal14.225.0+10.8
BuseyBench SVGPerception & media › Generation4.35.4+10.5
Ramp SWE-Bench v2Software › Software engineering60.851.8−9.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine26.718.3−8.3
Terminal-Bench 4.0Software › Software engineering31.839.4+7.6
Terminal-Bench-Science 0.1Science › Multidisciplinary12.920.0+7.1
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering33.840.9+7.1
LMArena Code Arena WebDevSoftware › Frontend & design16591608−6.7
AA-Briefcase v1.1 EloProfessional › General knowledge work14161526+6.0
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations24.029.8+5.8
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work17.222.8+5.6
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge20.825.1+4.3
Artificial Analysis Engineering IndexScience › Engineering37.641.8+4.2
Vibe Code Bench v1.1Software › Frontend & design87.783.7−4.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal18.822.7+3.9
Legal Research BenchProfessional › Legal45.249.0+3.8
Vals Legal IndexProfessional › Legal71.975.7+3.8
APEX-SWESoftware › Software engineering49.152.9+3.8
APEX-Agents 1.1Professional › General knowledge work79.475.7−3.7
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations58.562.1+3.6
Tax Agent BenchProfessional › Finance, accounting & tax74.971.4−3.6
Artificial Analysis Intelligence Index v5.0.1Composite indexes32.636.0+3.4
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge1.57.7+3.1
GDP.xlsxProfessional › General knowledge work18.415.4−3.0
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax69.272.2+3.0
Artificial Analysis Coding Agent Index v2.1Software › Software engineering36.839.7+2.9
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents1.16.6+2.8
Artificial Analysis Economics IndexComposite indexes38.541.0+2.5
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax42.244.6+2.4
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents2.2-2.0−2.1
APEX-1 (Mercor, mean score)Professional › General knowledge work68.370.4+2.1
Artificial Analysis Progress Index v1Composite indexes324329+1.8
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering28.527.2−1.3
LMArena Agent Arena Steering BurdenTool & computer use › General agents2.34.8+1.2
GBENCH One-shot Coding GScoreGames55.456.4+0.9
Finance Agent v2.1Professional › Finance, accounting & tax62.963.8+0.9
Vals Index v2.2Composite indexes61.360.5−0.8
Artificial Analysis Legal IndexProfessional › Legal37.738.5+0.7
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents-1.5-2.7−0.6
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work49.248.7−0.5
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling-0.10.9+0.5
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations68.568.1−0.4
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents0.81.0+0.4
Artificial Analysis Healthcare & Medical IndexHealth & medicine41.341.30.0
Vals Finance IndexProfessional › Finance, accounting & tax60.960.90.0
Data: Artificial Analysis, Vals AI, Arena and 6 othersatlas.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…