Atlas

Compare Models

Gemini 4 Neon vs GPT-6.2 Sol

Gemini 4 Neon leads by 2.0 capability points (≈95% interval −0.9 to +4.9): statistically tied · P(higher) ≈ 91%

Wins 52 of 59 shared benchmarks (7 lost)

Capability vs. relative cost
Gemini 4 NeonGPT-6.2 SolClosed weightsOpen weights

Benchmark wins52 – 7 · 59 shared

Comparing matching setups only (31 benchmarks excluded) · suites counted once
← Gemini 4 NeonGPT-6.2 Sol →
Data: Artificial Analysis, Vals AI, Surge AI and 7 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.2 Sol minus Gemini 4 Neon.

EnterpriseBench: CoreCraftTool & computer use › Customer service & operations77.436.4−41.0
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine47.217.2−30.0
SimpleBenchKnowledge & reasoning › General knowledge94.469.9−24.5
DAYJOB: AccountingProfessional › Finance, accounting & tax49.627.6−22.0
GDP.xlsxProfessional › General knowledge work43.122.3−20.8
Legal Research BenchProfessional › Legal56.337.0−19.2
Humanity's Sixth SensePerception & media › Spatial & visual reasoning75.956.8−19.0
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations52.334.5−17.8
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax50.532.8−17.7
Tax Agent BenchProfessional › Finance, accounting & tax79.862.9−16.9
GraphWalks BFS: 256K to 1M AggregateKnowledge & reasoning › Reading & long context95.280.1−15.1
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations56.341.3−15.1
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents18.411.3−13.8
Vals Legal IndexProfessional › Legal78.466.7−11.7
Vals Finance IndexProfessional › Finance, accounting & tax68.356.7−11.7
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents44.523.3−10.6
Finance Agent v2.1Professional › Finance, accounting & tax68.658.9−9.6
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering75.885.3+9.6
Artificial Analysis Legal IndexProfessional › Legal63.855.0−8.7
MysteryMechanismMath45.036.9−8.1
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations79.371.3−8.1
DAYJOB: TaxProfessional › Finance, accounting & tax12.84.8−8.0
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax65.157.2−7.8
HLAB-Vals-1.1Professional › Legal19.612.1−7.5
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge71.364.0−7.3
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge54.840.4−7.2
LiveBench 2.0 (2026-11-06 release)Composite indexes87.480.3−7.1
Artificial Analysis Healthcare & Medical IndexHealth & medicine60.153.3−6.8
AA-Briefcase v1.1 EloProfessional › General knowledge work16991580−6.5
DAYJOB: Law (November 2026 release)Professional › Legal26.232.7+6.5
Vibe Code Bench v1.1Software › Frontend & design93.787.6−6.1
Artificial Analysis Engineering IndexScience › Engineering66.861.2−5.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine31.737.2+5.6
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-6.04.3+5.1
Vals Science IndexScience › Multidisciplinary62.657.8−4.9
Vals Index v2.2Composite indexes70.966.1−4.8
GDP.pdf (November 2026 rescoring)Professional › General knowledge work53.448.8−4.6
Artificial Analysis Economics IndexComposite indexes61.957.4−4.6
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work63.859.4−4.4
Terminal-Bench Science 0.1 (Vals)Science › Multidisciplinary62.958.6−4.3
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents19.210.9−4.1
Artificial Analysis Intelligence Index v5.0.1Composite indexes61.457.3−4.1
RiemannBench (November 2026 rescoring)Math84.080.0−4.0
LMArena Agent Arena Steering BurdenTool & computer use › General agents21.213.4−3.9
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following72.969.5−3.4
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work33.430.0−3.4
Terminal-Bench-Science 0.1Science › Multidisciplinary70.066.7−3.3
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering61.658.6−3.0
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work75.873.0−2.8
BioMysteryBench (Vals Terminus 2)Science › Life sciences80.077.8−2.2
sudo L7 (October 2026 rescoring)Software › Software engineering38.336.1−2.2
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.574.3−2.2
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering44.442.8−1.7
ChartographyPerception & media › Documents & charts73.575.1+1.6
LVBenchPerception & media › Video understanding89.087.6−1.4
SWE-sweepSoftware › Software engineering12.811.6−1.2
Terminal-Bench 4.0Software › Software engineering61.660.6−1.0
CritPt v2 (agentic, avg@5)Science › Physical sciences78.078.3+0.3
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.01.2+0.1
Data: Artificial Analysis, Vals AI, Surge AI and 7 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…