Atlas

Compare Models

Claude Opus 5.5 vs Gemini 4 Neon

Gemini 4 Neon leads by 1.3 capability points (≈95% interval −1.6 to +4.3): statistically tied · P(higher) ≈ 82%

Wins 37 of 51 shared benchmarks (11 lost, 3 tied)

Capability vs. relative cost
Claude Opus 5.5Gemini 4 NeonClosed weightsOpen weights

Benchmark wins11 – 37 · 51 shared, 3 tied

Comparing matching setups only (15 benchmarks excluded) · suites counted once
← Claude Opus 5.5Gemini 4 Neon →
Data: Surge AI, Arena, Artificial Analysis and 12 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Neon minus Claude Opus 5.5.

Humanity's Sixth SensePerception & media › Spatial & visual reasoning44.675.9+31.3
SWE-sweepSoftware › Software engineering10.139.0+28.9
Bug Hunt BenchSoftware › Software engineering39.760.0+20.3
SWE-Marathon v1.2Software › Software engineering81.362.5−18.7
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge54.671.3+16.7
Chess Puzzles (task version 2.0)Games67.083.0+16.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal10.826.6+15.8
DAYJOB: AccountingProfessional › Finance, accounting & tax34.449.6+15.2
Mystery Game PuzzlesGames71.085.0+14.0
GDP.xlsxProfessional › General knowledge work30.343.1+12.8
HLAB-Vals-1.1Professional › Legal9.219.6+10.4
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents14.0-6.0−10.0
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following64.472.9+8.5
Terminal-Bench-Science 0.1Science › Multidisciplinary61.970.0+8.1
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax42.550.5+8.0
sudo L7 (October 2026 rescoring)Software › Software engineering46.138.3−7.8
APEX-Agents 1.1Professional › General knowledge work81.388.9+7.6
CursorBench 4.1Software › Software engineering58.165.4+7.3
LMArena Code Arena WebDevSoftware › Frontend & design18181764−7.2
SimpleBenchKnowledge & reasoning › General knowledge88.494.4+6.0
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning83.389.2+5.8
APEX-Accounting 1.1Professional › Finance, accounting & tax65.570.8+5.3
GDP.pdf (November 2026 rescoring)Professional › General knowledge work48.853.4+4.6
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work28.833.4+4.6
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations55.059.5+4.5
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering59.663.6+4.0
CritPt v2 (agentic, avg@5)Science › Physical sciences81.477.4−4.0
LiveBench 2.0 (2026-11-06 release)Composite indexes83.487.4+4.0
RiemannBench (November 2026 rescoring)Math80.084.0+4.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine27.831.7+3.9
CWE-bench v2 Pass@1Software › Cybersecurity67.170.7+3.6
DAYJOB: TaxProfessional › Finance, accounting & tax9.212.8+3.6
AA-Briefcase v1.1 EloProfessional › General knowledge work17601699−3.4
Chartography (November 2026 rescoring)Perception & media › Documents & charts89.686.6−3.0
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations74.477.4+3.0
SimpleQA Verified (Epoch anti-abstention)Knowledge & reasoning › General knowledge72.275.2+3.0
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents38.644.5+2.9
Vals RSI Index v1.1 (four tasks)Software › Data & ML37.334.4−2.9
FrontierMath Tier 4 (v2) - PrivateMath95.097.6+2.6
DAYJOB: Law (November 2026 release)Professional › Legal23.726.2+2.5
LMArena Text Arena EloLanguage & writing › General assistance15071518+2.0
LMArena Agent Arena Steering BurdenTool & computer use › General agents18.021.2+1.6
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine48.347.2−1.1
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering43.844.4+0.7
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax75.876.5+0.6
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents18.618.4−0.4
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents19.219.20.0
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.01.00.0
FrontierMath Tiers 1-3 (v2) - PrivateMath91.291.20.0
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations52.352.30.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
Data: Surge AI, Arena, Artificial Analysis and 12 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…