Atlas

Compare Models

Gemini 4 Hydrogen vs GPT-6.2 Sol

GPT-6.2 Sol leads by 1.3 capability points (≈95% interval −1.4 to +3.9): statistically tied · P(higher) ≈ 82%

Gemini 4 Hydrogen wins 36 of 62 shared benchmarks (26 lost)

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

Benchmark wins36 – 26 · 62 shared

Comparing matching setups only (33 benchmarks excluded) · suites counted once
← Gemini 4 HydrogenGPT-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 4 Hydrogen.

KradleBenchGames43.967.9+24.0
DAYJOB: Law (November 2026 release)Professional › Legal9.732.7+23.0
SimpleBenchKnowledge & reasoning › General knowledge91.769.9−21.8
GDP.xlsxProfessional › General knowledge work40.922.3−18.6
Terminal-Bench-Science 0.1Science › Multidisciplinary49.566.7+17.1
Legal Research BenchProfessional › Legal53.837.0−16.8
Artificial Analysis Intelligence Index v5.0.1Composite indexes41.257.3+16.0
Tax Agent BenchProfessional › Finance, accounting & tax78.462.9−15.5
TaxCalcBench TY25Professional › Finance, accounting & tax44.032.0−12.0
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations44.634.5−10.1
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine27.237.2+10.0
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations51.141.3−9.9
Professional Reasoning Benchmark - LegalProfessional › Legal53.143.3−9.7
Vals Finance IndexProfessional › Finance, accounting & tax66.156.7−9.5
MultiChallengeLanguage & writing › Conversation80.672.0−8.6
ChartographyPerception & media › Documents & charts66.675.1+8.5
Vals Legal IndexProfessional › Legal73.666.7−7.0
DrugDiscoveryBenchScience › Life sciences69.162.2−6.9
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine24.017.2−6.8
ARC-AGI-2Knowledge & reasoning › Abstract reasoning92.585.8−6.7
Terminal-Bench 4.0Software › Software engineering54.060.6+6.6
Humanity's Sixth SensePerception & media › Spatial & visual reasoning63.456.8−6.5
Finance Agent v2.1Professional › Finance, accounting & tax65.158.9−6.2
DAYJOB: AccountingProfessional › Finance, accounting & tax33.627.6−6.0
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax38.532.8−5.7
GDP.pdf (November 2026 rescoring)Professional › General knowledge work54.248.8−5.4
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge58.864.0+5.2
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering54.058.6+4.5
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-4.54.3+4.4
sudo L7 (October 2026 rescoring)Software › Software engineering31.736.1+4.4
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents13.511.3−4.3
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following73.769.5−4.2
TxBench Preclinical Pharmacology V1 Clean RerunsScience › Life sciences61.365.3+4.0
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering39.042.8+3.8
Dual-Arm Sim (task completion)Tool & computer use › Robotics0.03.8+3.8
LiveBench 2.0 (2026-11-06 release)Composite indexes77.080.3+3.3
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations74.471.3−3.2
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents29.323.3−3.0
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge46.340.4−2.9
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work32.830.0−2.8
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering82.785.3+2.7
Vals Index v2.2Composite indexes68.566.1−2.4
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents15.710.9−2.4
Artificial Analysis Economics IndexComposite indexes55.057.4+2.3
Gray Swan IPI Attack Success (k=15)Safety & alignment0.02.3−2.3
CritPt v2 (agentic, avg@5)Science › Physical sciences80.678.3−2.3
LMArena Agent Arena Steering BurdenTool & computer use › General agents17.413.4−2.0
Artificial Analysis Legal IndexProfessional › Legal56.755.0−1.7
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work57.759.4+1.7
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering70.872.4+1.6
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax75.874.3−1.5
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work71.673.0+1.4
DAYJOB: TaxProfessional › Finance, accounting & tax3.64.8+1.2
Artificial Analysis Engineering IndexScience › Engineering60.161.2+1.1
ARC-AGI-1 Semi-PrivateKnowledge & reasoning › Abstract reasoning98.597.5−1.0
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax83.284.2+1.0
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax58.157.2−0.8
HLAB-Vals-1.1Professional › Legal11.312.1+0.8
AA-Briefcase v1.1 EloProfessional › General knowledge work15691580+0.6
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.71.2+0.2
Vibe Code Bench v1.1Software › Frontend & design87.887.6−0.2
Artificial Analysis Healthcare & Medical IndexHealth & medicine53.353.3−0.1
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…