Atlas

Compare Models

Gemini 4 Neon vs GPT-6.1 Sol

Gemini 4 Neon leads by 2.8 capability points (≈95% interval −0.1 to +5.7): statistically tied · P(higher) ≈ 97%

Wins 54 of 63 shared benchmarks (8 lost, 1 tied)

Capability vs. relative cost
Gemini 4 NeonGPT-6.1 SolClosed weightsOpen weights

Benchmark wins54 – 8 · 63 shared, 1 tied

Comparing matching setups only (14 benchmarks excluded) · suites counted once
← Gemini 4 NeonGPT-6.1 Sol →
Data: Vals AI, Artificial Analysis, Surge AI and 10 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.1 Sol minus Gemini 4 Neon.

Humanity's Sixth SensePerception & media › Spatial & visual reasoning75.946.6−29.3
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations77.450.3−27.1
DAYJOB: AccountingProfessional › Finance, accounting & tax49.625.2−24.4
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations52.328.6−23.7
Vending-Bench 2Business simulations1684312861−23.6
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine47.225.2−22.0
GDP.xlsxProfessional › General knowledge work43.122.0−21.1
Terminal-Bench-Science 0.1Science › Multidisciplinary70.050.5−19.5
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal26.68.8−17.8
Legal Research BenchProfessional › Legal56.338.5−17.8
Tax Agent BenchProfessional › Finance, accounting & tax79.862.3−17.5
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax50.533.5−17.0
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations59.543.0−16.5
APEX-Agents 1.1Professional › General knowledge work88.973.2−15.7
Chess Puzzles (task version 2.0)Games83.068.0−15.0
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents18.411.3−13.9
HLAB-Vals-1.1Professional › Legal19.66.3−13.3
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations79.366.6−12.8
Vals Finance IndexProfessional › Finance, accounting & tax68.355.8−12.5
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge71.358.8−12.5
SimpleBenchKnowledge & reasoning › General knowledge94.482.9−11.5
AA-Briefcase v1.1 EloProfessional › General knowledge work16991504−10.7
APEX-1 (Mercor, mean score)Professional › General knowledge work79.469.3−10.1
Terminal-Bench Science 0.1 (Vals)Science › Multidisciplinary62.952.9−10.0
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents44.524.5−10.0
Vals Legal IndexProfessional › Legal78.468.7−9.6
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering63.654.0−9.6
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning89.280.0−9.2
DAYJOB: TaxProfessional › Finance, accounting & tax12.84.0−8.8
MKA (Artificial Analysis)Professional › General knowledge work56.547.7−8.8
Vals Index v2.2Composite indexes70.963.9−6.9
Artificial Analysis Intelligence Index v5.0.1Composite indexes61.454.6−6.8
Finance Agent v2.1Professional › Finance, accounting & tax68.661.8−6.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine31.738.3+6.7
APEX-Accounting 1.1Professional › Finance, accounting & tax70.864.3−6.5
LiveBench 2.0 (2026-11-06 release)Composite indexes87.481.2−6.3
LMArena Text Arena EloLanguage & writing › General assistance15181483−6.2
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge54.842.4−6.2
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering44.438.5−5.9
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-6.05.1+5.5
DAYJOB: Law (November 2026 release)Professional › Legal26.231.6+5.4
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.571.1−5.4
CritPt v2 (agentic, avg@5)Science › Physical sciences77.482.6+5.1
Mystery Game PuzzlesGames85.080.0−5.0
RiemannBench (November 2026 rescoring)Math84.079.2−4.8
Vibe Code Bench v1.1Software › Frontend & design93.788.9−4.8
LMArena Agent Arena Steering BurdenTool & computer use › General agents21.211.8−4.7
sudo L7 (October 2026 rescoring)Software › Software engineering38.333.9−4.4
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents19.211.6−3.8
Terminal-Bench 4.0Software › Software engineering61.658.2−3.4
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work75.872.5−3.3
Vals Science IndexScience › Multidisciplinary62.659.6−3.0
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following72.970.1−2.8
FrontierMath Tiers 1-3 (v2) - PrivateMath91.293.7+2.5
FrontierMath Tier 4 (v2) - PrivateMath97.6100.0+2.4
GDP.pdf (November 2026 rescoring)Professional › General knowledge work53.451.6−1.8
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work33.432.0−1.4
MysteryMechanismMath45.046.4+1.4
SimpleQA Verified (Epoch anti-abstention)Knowledge & reasoning › General knowledge75.273.9−1.3
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering75.875.1−0.6
BioMysteryBench (Vals Terminus 2)Science › Life sciences80.079.6−0.4
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.01.2+0.1
OTIS Mock AIME 2024-2025Math100.0100.00.0
Data: Vals AI, Artificial Analysis, Surge AI and 10 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…