Atlas

Compare Models

Gemini 3.8 Flash vs GPT-6.2 Sol

GPT-6.2 Sol leads by 6.3 capability points (≈95% interval +3.8 to +8.8): a clear lead over Gemini 3.8 Flash · P(higher) ≈ 100%

Wins 61 of 79 shared benchmarks (17 lost, 1 tied)

Capability vs. relative cost
Gemini 3.8 FlashGPT-6.2 SolClosed weightsOpen weights

Benchmark wins17 – 61 · 79 shared, 1 tied

Comparing matching setups only (10 benchmarks excluded) · suites counted once
← Gemini 3.8 FlashGPT-6.2 Sol →
Data: Artificial Analysis, Vals AI, Surge AI and 15 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 3.8 Flash.

Terminal-Bench-Science 0.1Science › Multidisciplinary10.067.6+57.6
AA-PersonalityBench Slop indexLanguage & writing › Style & personality71.415.4−56.6
ARC-AGI-3 (Semi-Private)Games35.090.1+55.1
Terminal-Bench Science 0.1 (Vals)Science › Multidisciplinary8.658.6+50.0
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering38.285.3+47.2
AA-PersonalityBench Informativeness indexLanguage & writing › Style & personality81.037.7−43.7
KradleBenchGames24.367.9+43.6
Terminal-Bench 4.0Software › Software engineering19.760.6+40.9
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering19.758.6+38.9
AA-3D v1.0 EloSoftware › Frontend & design9741259+38.8
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge29.164.0+34.9
DeepsecBench-AA (Artificial Analysis)Software › Cybersecurity14.549.3+34.8
ChartographyPerception & media › Documents & charts40.975.1+34.2
AA-PersonalityBench Transparency indexLanguage & writing › Style & personality23.856.9+33.4
DAYJOB: Law (November 2026 release)Professional › Legal0.032.7+32.7
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax4.532.8+28.3
AA-PersonalityBench Organization indexLanguage & writing › Style & personality73.345.7−27.9
Artificial Analysis Intelligence Index v5.0.1Composite indexes30.857.3+26.4
TaxCalcBench TY25Professional › Finance, accounting & tax6.032.0+26.0
sudo L7 (October 2026 rescoring)Software › Software engineering10.636.1+25.5
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering47.072.4+25.4
AA-PersonalityBench Expressiveness indexLanguage & writing › Style & personality79.155.3−24.0
RiemannBench (November 2026 rescoring)Math56.880.0+23.2
LiveBench 2.0 (2026-11-06 release)Composite indexes57.380.3+23.1
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations58.536.4−22.1
Vals Science IndexScience › Multidisciplinary35.857.8+22.0
VIALS-AA (pass^3)Science › Life sciences41.062.7+21.7
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine15.637.2+21.7
DAYJOB: AccountingProfessional › Finance, accounting & tax6.027.6+21.6
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering18.740.2+21.5
AA-PersonalityBench Imagination indexLanguage & writing › Style & personality41.060.1+19.3
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents1.511.3+19.1
AA-Briefcase v1.1 EloProfessional › General knowledge work12041541+18.5
Blueprint-Bench 2Perception & media › Spatial & visual reasoning0.40.6+18.4
VIALS-AA (average of 3)Science › Life sciences52.068.9+17.0
BioMysteryBench (Vals Terminus 2)Science › Life sciences62.277.8+15.6
Humanity's Sixth SensePerception & media › Spatial & visual reasoning41.656.8+15.2
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine2.017.2+15.2
CritPt v2 (agentic, avg@5)Science › Physical sciences65.177.7+12.6
SimpleBenchKnowledge & reasoning › General knowledge82.469.9−12.5
Artificial Analysis Cyber Index v1Software › Cybersecurity25.137.0+11.9
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations29.741.3+11.6
TxBench Preclinical Pharmacology V1 Clean RerunsScience › Life sciences54.765.3+10.7
VIALS-AA (pass@3)Science › Life sciences64.074.5+10.6
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations60.971.3+10.4
Vals Index v2.2Composite indexes56.066.1+10.1
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents4.023.3+9.6
CWE-Bench-AA (Artificial Analysis)Software › Cybersecurity52.561.7+9.2
Vibe Code Bench v1.1Software › Frontend & design78.787.6+9.0
CyberGym-E2E-AA (Artificial Analysis)Software › Cybersecurity8.40.0−8.4
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge23.740.4+8.3
LMArena Agent Arena Steering BurdenTool & computer use › General agents-2.813.4+8.1
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations26.534.5+8.0
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work22.830.0+7.2
Finance Agent v2.1Professional › Finance, accounting & tax65.858.9−6.8
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following63.269.5+6.3
AA-AnalystAgent (pass^5)Professional › General knowledge work46.352.5+6.3
AA-PersonalityBench Conviction indexLanguage & writing › Style & personality84.879.2−5.6
Professional Reasoning Benchmark - LegalProfessional › Legal48.943.3−5.6
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents1.010.9+5.0
GDP.pdf (November 2026 rescoring)Professional › General knowledge work44.248.8+4.6
DAYJOB: TaxProfessional › Finance, accounting & tax0.84.8+4.0
GDP.xlsxProfessional › General knowledge work18.322.3+4.0
Tax Agent BenchProfessional › Finance, accounting & tax66.862.9−3.9
Vals Finance IndexProfessional › Finance, accounting & tax60.556.7−3.8
Dual-Arm Sim (task completion)Tool & computer use › Robotics0.03.8+3.8
ARC-AGI-2Knowledge & reasoning › Abstract reasoning89.285.8−3.3
Gray Swan IPI Attack Success (k=15)Safety & alignment5.52.3+3.2
SWE Atlas v1.1 - OverallSoftware › Software engineering57.360.5+3.1
Vals Legal IndexProfessional › Legal69.866.7−3.1
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents8.94.3−2.3
Legal Research BenchProfessional › Legal38.937.0−1.9
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax72.974.3+1.4
ARC-AGI-1 Semi-PrivateKnowledge & reasoning › Abstract reasoning98.597.5−1.0
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax83.284.2+1.0
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.01.2+0.6
MysteryMechanismMath36.536.9+0.5
AA-PersonalityBench Warmth indexLanguage & writing › Style & personality60.060.1+0.1
HLAB-Vals-1.1Professional › Legal12.112.10.0
Data: Artificial Analysis, Vals AI, Surge AI and 15 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…