Atlas

Compare Models

Gemini 4 Neon vs Muse Wonder

Gemini 4 Neon leads by 5.5 capability points (≈95% interval +2.8 to +8.3): a clear lead over Muse Wonder · P(higher) ≈ 100%

Wins 67 of 70 shared benchmarks (3 lost)

Capability vs. relative cost
Gemini 4 NeonMuse WonderClosed weightsOpen weights

Benchmark wins67 – 3 · 70 shared

Comparing matching setups only (2 benchmarks excluded) · suites counted once
← Gemini 4 NeonMuse Wonder →
Data: Artificial Analysis, Vals AI, Surge AI and 11 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Wonder minus Gemini 4 Neon.

Vending-Bench 2Business simulations168439432−43.9
Terminal-Bench-Science 0.1Science › Multidisciplinary65.721.9−43.8
Terminal-Bench Science 0.1 (Vals)Science › Multidisciplinary62.922.9−40.0
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning89.254.2−35.0
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents18.40.6−34.8
Mystery Game PuzzlesGames85.051.0−34.0
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge71.339.0−32.3
Humanity's Sixth SensePerception & media › Spatial & visual reasoning75.943.7−32.2
Bug Hunt BenchSoftware › Software engineering60.030.2−29.8
FrontierMath Tier 4 (v2) - PrivateMath97.670.7−26.8
RiemannBench (November 2026 rescoring)Math84.059.2−24.8
SWE-Marathon v1.2Software › Software engineering62.538.1−24.4
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents44.5-3.0−23.7
Vals Science IndexScience › Multidisciplinary62.639.1−23.5
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering61.638.4−23.2
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations52.329.5−22.8
sudo L7 (October 2026 rescoring)Software › Software engineering38.316.1−22.2
GDP.xlsxProfessional › General knowledge work43.121.8−21.3
DAYJOB: AccountingProfessional › Finance, accounting & tax49.630.8−18.8
Artificial Analysis Intelligence Index v5.0.1Composite indexes61.443.6−17.8
MysteryMechanismMath45.027.5−17.6
Artificial Analysis Engineering IndexScience › Engineering66.849.6−17.3
Artificial Analysis Legal IndexProfessional › Legal63.846.9−16.9
LiveBench 2.0 (2026-11-06 release)Composite indexes87.470.7−16.7
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering43.326.6−16.7
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine47.231.2−16.0
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations77.461.5−15.9
LMArena Code Arena WebDevSoftware › Frontend & design17641648−15.4
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge54.825.1−14.8
Artificial Analysis Coding Agent Index v2.1Software › Software engineering58.644.5−14.2
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax65.151.1−13.9
BioMysteryBench (Vals Terminus 2)Science › Life sciences80.067.0−13.0
Artificial Analysis Economics IndexComposite indexes61.949.5−12.4
Terminal-Bench 4.0Software › Software engineering61.649.5−12.1
APEX-Agents 1.1Professional › General knowledge work88.977.0−11.9
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax50.538.8−11.7
Artificial Analysis Healthcare & Medical IndexHealth & medicine60.148.7−11.4
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations79.368.1−11.2
Vibe Code Bench v1.1Software › Frontend & design93.783.1−10.6
LMArena Agent Arena Steering BurdenTool & computer use › General agents21.21.5−9.8
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine31.722.2−9.4
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work63.854.5−9.3
FrontierMath Tiers 1-3 (v2) - PrivateMath91.282.1−9.1
MKA (Artificial Analysis)Professional › General knowledge work56.547.5−9.0
Tax Agent BenchProfessional › Finance, accounting & tax79.871.2−8.6
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents19.22.5−8.4
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations59.551.9−7.6
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work75.868.4−7.4
Vals Finance IndexProfessional › Finance, accounting & tax68.361.2−7.1
DAYJOB: TaxProfessional › Finance, accounting & tax12.86.0−6.8
Vals Index v2.2Composite indexes70.964.9−6.0
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering75.869.8−6.0
Legal Research BenchProfessional › Legal56.350.5−5.8
CritPt v2 (agentic, avg@5)Science › Physical sciences76.670.9−5.7
DAYJOB: Law (November 2026 release)Professional › Legal26.220.6−5.6
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following72.967.9−5.0
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-6.03.4+4.7
CWE-bench v2 Pass@1Software › Cybersecurity70.766.1−4.6
HLAB-Vals-1.1Professional › Legal19.623.8+4.2
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.572.5−4.0
APEX-Accounting 1.1Professional › Finance, accounting & tax70.867.1−3.7
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal26.623.0−3.6
Vals Legal IndexProfessional › Legal78.475.3−3.1
Finance Agent v2.1Professional › Finance, accounting & tax68.665.5−3.0
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work33.430.6−2.8
GDP.pdf (November 2026 rescoring)Professional › General knowledge work53.454.8+1.4
LMArena Text Arena EloLanguage & writing › General assistance15181511−1.4
AA-Briefcase v1.1 EloProfessional › General knowledge work16321610−1.2
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.0-0.2−0.6
OTIS Mock AIME 2024-2025Math100.099.4−0.6
Data: Artificial Analysis, Vals AI, Surge AI and 11 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…