Atlas

Compare Models

Gemini 4 Carbon vs Gemini 4 Neon

Gemini 4 Neon leads by 1.3 capability points (≈95% interval −1.7 to +4.2): statistically tied · P(higher) ≈ 80%

Wins 56 of 78 shared benchmarks (20 lost, 2 tied)

Capability vs. relative cost
Gemini 4 CarbonGemini 4 NeonClosed weightsOpen weights

Benchmark wins20 – 56 · 78 shared, 2 tied

Comparing matching setups only · suites counted once
← Gemini 4 CarbonGemini 4 Neon →
Data: Artificial Analysis, Surge AI, Vals AI and 14 othersatlas.village.dev

Margins

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

DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax23.850.5+26.7
DAYJOB: AccountingProfessional › Finance, accounting & tax25.249.6+24.4
DAYJOB: Law (November 2026 release)Professional › Legal13.826.2+12.4
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine19.431.7+12.2
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine35.247.2+12.0
Mystery Game PuzzlesGames73.085.0+12.0
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning77.589.2+11.7
DAYJOB: TaxProfessional › Finance, accounting & tax4.812.8+8.0
sudo L7 (October 2026 rescoring)Software › Software engineering30.638.3+7.7
Terminal-Bench-Science 0.1Science › Multidisciplinary62.454.8−7.6
Vals RSI Index v1.1 (four tasks)Software › Data & ML27.734.4+6.7
ExploitGym - release-reported percentage, cohort unspecifiedSoftware › Cybersecurity54.848.3−6.5
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering60.654.5−6.1
Chess Puzzles (task version 2.0)Games78.083.0+5.0
Humanity's Sixth SensePerception & media › Spatial & visual reasoning71.175.9+4.8
Vibe Code Bench v1.1Software › Frontend & design89.193.7+4.6
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents16.118.4+4.5
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering80.175.8−4.3
CritPt v2 (agentic, avg@5)Science › Physical sciences82.378.0−4.3
GDP.xlsxProfessional › General knowledge work38.943.1+4.2
SWE-Marathon v1.2Software › Software engineering66.762.5−4.2
LiveBench 2.0 (2026-11-06 release)Composite indexes83.387.4+4.1
RiemannBench (November 2026 rescoring)Math80.084.0+4.0
CursorBench 4.1Software › Software engineering61.565.4+3.9
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations55.759.5+3.8
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work29.633.4+3.8
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal23.026.6+3.6
CWE-bench v2 Pass@1Software › Cybersecurity67.370.7+3.4
Tax Agent BenchProfessional › Finance, accounting & tax76.479.8+3.4
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations53.056.3+3.4
Artificial Analysis Healthcare & Medical IndexHealth & medicine56.860.1+3.3
SWE-sweepSoftware › Software engineering9.612.8+3.2
Artificial Analysis Coding Agent Index v2.1Software › Software engineering55.558.6+3.2
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations80.577.4−3.1
GDP.pdf (November 2026 rescoring)Professional › General knowledge work56.453.4−3.0
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work60.963.8+2.8
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering40.643.3+2.7
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents39.444.5+2.5
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations49.852.3+2.5
SimpleQA Verified (Epoch anti-abstention)Knowledge & reasoning › General knowledge77.775.2−2.5
FrontierMath Tier 4 (v2) - PrivateMath95.197.6+2.4
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax62.665.1+2.4
Vals Legal IndexProfessional › Legal75.978.4+2.4
Artificial Analysis Economics IndexComposite indexes59.661.9+2.3
SEC-Bench Pro - release-reported cohortSoftware › Cybersecurity88.290.5+2.3
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge69.171.3+2.2
HLAB-Vals-1.1Professional › Legal17.519.6+2.1
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following70.972.9+2.0
SimpleBenchKnowledge & reasoning › General knowledge92.694.4+1.8
Artificial Analysis Legal IndexProfessional › Legal62.063.8+1.8
LMArena Agent Arena Steering BurdenTool & computer use › General agents17.921.2+1.6
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work77.475.8−1.6
APEX-Agents 1.1Professional › General knowledge work90.488.8−1.6
Artificial Analysis Engineering IndexScience › Engineering65.366.8+1.5
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations77.879.3+1.5
Terminal-Bench 4.0Software › Software engineering63.161.6−1.5
Finance Agent v2.1Professional › Finance, accounting & tax67.268.6+1.4
Artificial Analysis Intelligence Index v5.0.1Composite indexes60.061.4+1.4
LMArena Text Arena EloLanguage & writing › General assistance15261518−1.3
Vals Finance IndexProfessional › Finance, accounting & tax67.068.3+1.3
AA-Briefcase v1.1 EloProfessional › General knowledge work16111632+1.2
APEX-1 (Mercor, mean score)Professional › General knowledge work80.078.9−1.1
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents17.219.2+1.0
ChartographyPerception & media › Documents & charts72.573.5+1.0
GraphWalks BFS: 256K to 1M AggregateKnowledge & reasoning › Reading & long context96.295.2−1.0
FrontierMath Tiers 1-3 (v2) - PrivateMath90.591.2+0.7
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge53.554.8+0.6
LMArena Code Arena WebDevSoftware › Frontend & design17691764−0.6
Legal Research BenchProfessional › Legal55.856.3+0.5
Bug Hunt BenchSoftware › Software engineering54.654.3−0.3
GPQA DiamondScience › Multidisciplinary96.596.3−0.3
MKA (Artificial Analysis)Professional › General knowledge work54.354.6+0.3
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-5.7-6.0−0.1
APEX-Accounting 1.1Professional › Finance, accounting & tax70.770.8+0.1
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.81.0+0.1
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.576.50.0
Gray Swan IPI Attack Success (k=15)Safety & alignment0.00.00.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
Data: Artificial Analysis, Surge AI, Vals AI and 14 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…