Atlas

Compare Models

Gemini 4 Carbon vs GPT-6.2 Sol

Gemini 4 Carbon leads by 0.8 capability points (≈95% interval −2.0 to +3.5): statistically tied · P(higher) ≈ 71%

Wins 52 of 70 shared benchmarks (16 lost, 2 tied)

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

Benchmark wins52 – 16 · 70 shared, 2 tied

Comparing matching setups only (38 benchmarks excluded) · suites counted once
← Gemini 4 CarbonGPT-6.2 Sol →
Data: Artificial Analysis, Surge AI, Vals 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 4 Carbon.

KradleBenchGames22.067.9+46.0
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations80.536.4−44.1
SimpleBenchKnowledge & reasoning › General knowledge92.669.9−22.7
DAYJOB: Law (November 2026 release)Professional › Legal13.832.7+18.9
Legal Research BenchProfessional › Legal55.837.0−18.8
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine35.217.2−18.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine19.437.2+17.8
GDP.xlsxProfessional › General knowledge work38.922.3−16.6
GraphWalks BFS: 256K to 1M AggregateKnowledge & reasoning › Reading & long context96.280.1−16.1
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations49.834.5−15.3
Humanity's Sixth SensePerception & media › Spatial & visual reasoning71.156.8−14.3
MultiChallengeLanguage & writing › Conversation85.972.0−13.9
Tax Agent BenchProfessional › Finance, accounting & tax76.462.9−13.5
DeepsecBench — Revalidated V1 (233 issues)Software › Cybersecurity35.247.3+12.2
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations53.041.3−11.7
SWE Atlas v1.1 - OverallSoftware › Software engineering71.560.5−11.0
Vals Finance IndexProfessional › Finance, accounting & tax67.056.7−10.4
Professional Reasoning Benchmark - LegalProfessional › Legal53.443.3−10.0
Dual-Arm Sim (task completion)Tool & computer use › Robotics13.83.8−10.0
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents16.111.3−9.3
DrugDiscoveryBenchScience › Life sciences71.562.2−9.3
Vals Legal IndexProfessional › Legal75.966.7−9.3
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax23.832.8+9.0
ARC-AGI-2Knowledge & reasoning › Abstract reasoning94.285.8−8.3
Finance Agent v2.1Professional › Finance, accounting & tax67.258.9−8.2
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents39.423.3−8.1
GDP.pdf (November 2026 rescoring)Professional › General knowledge work56.448.8−7.6
Artificial Analysis Legal IndexProfessional › Legal62.055.0−6.9
ARC-AGI-3 (Semi-Private)Games96.990.1−6.8
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge53.540.4−6.6
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations77.871.3−6.5
sudo L7 (October 2026 rescoring)Software › Software engineering30.636.1+5.5
HLAB-Vals-1.1Professional › Legal17.512.1−5.4
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax62.657.2−5.4
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering80.185.3+5.3
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge69.164.0−5.1
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-5.74.3+5.0
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering60.656.1−4.5
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work77.473.0−4.4
Artificial Analysis Engineering IndexScience › Engineering65.361.2−4.1
CritPt v2 (agentic, avg@5)Science › Physical sciences82.378.3−4.0
AA-Briefcase v1.1 EloProfessional › General knowledge work16111541−3.8
VIALS-AA (pass@3)Science › Life sciences70.874.5+3.7
Artificial Analysis Healthcare & Medical IndexHealth & medicine56.853.3−3.5
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents17.210.9−3.1
LiveBench 2.0 (2026-11-06 release)Composite indexes83.380.3−3.0
Artificial Analysis Intelligence Index v5.0.1Composite indexes60.057.3−2.8
ChartographyPerception & media › Documents & charts72.575.1+2.6
Terminal-Bench 4.0Software › Software engineering63.160.6−2.5
DAYJOB: AccountingProfessional › Finance, accounting & tax25.227.6+2.4
LMArena Agent Arena Steering BurdenTool & computer use › General agents17.913.4−2.3
Artificial Analysis Economics IndexComposite indexes59.657.4−2.3
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.574.3−2.2
SWE-sweepSoftware › Software engineering9.611.6+2.0
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering39.541.4+1.9
Terminal-Bench-Science 0.1Science › Multidisciplinary62.460.5−1.9
GDPval v2 (OpenAI internal)Professional › General knowledge work48.846.9−1.9
VIALS-AA (average of 3)Science › Life sciences67.368.9+1.7
Blueprint-Bench 2Perception & media › Spatial & visual reasoning0.60.6−1.6
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering74.072.4−1.6
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work60.959.4−1.5
Vibe Code Bench v1.1Software › Frontend & design89.187.6−1.4
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following70.969.5−1.4
VIALS-AA (pass^3)Science › Life sciences64.062.7−1.2
ARC-AGI-1 Semi-PrivateKnowledge & reasoning › Abstract reasoning98.597.5−1.0
TxBench Preclinical Pharmacology V1 Clean RerunsScience › Life sciences66.065.3−0.7
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work29.630.0+0.4
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.81.2+0.2
DAYJOB: TaxProfessional › Finance, accounting & tax4.84.80.0
RiemannBench (November 2026 rescoring)Math80.080.00.0
Data: Artificial Analysis, Surge AI, Vals 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…