Atlas

Compare Models

Gemini 4 Argon vs Gemini 4 Hydrogen

Gemini 4 Hydrogen leads by 0.1 capability points (≈95% interval −2.5 to +2.6): statistically tied · P(higher) ≈ 52%

Wins 55 of 85 shared benchmarks (27 lost, 3 tied)

Capability vs. relative cost
Gemini 4 ArgonGemini 4 HydrogenClosed weightsOpen weights

Benchmark wins27 – 55 · 85 shared, 3 tied

Comparing matching setups only (1 benchmark excluded) · suites counted once
← Gemini 4 ArgonGemini 4 Hydrogen →
Data: Artificial Analysis, Surge AI, Epoch AI and 24 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Hydrogen minus Gemini 4 Argon.

Terminal-Bench-Science 0.1Science › Multidisciplinary48.615.2−33.3
TxBench-Antibody-Development V1 (60 evaluations)Science › Life sciences31.160.6+29.5
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering55.129.3−25.8
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge42.558.8+16.3
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering54.870.8+16.0
HLAB-Vals-1.1Professional › Legal27.111.3−15.8
DAYJOB: AccountingProfessional › Finance, accounting & tax18.033.6+15.6
sudo L7 (October 2026 rescoring)Software › Software engineering16.131.7+15.6
KradleBenchGames28.643.9+15.3
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning72.557.5−15.0
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine13.927.2+13.3
FrontierCode 2.0 - Pass RateSoftware › Software engineering30.643.2+12.6
Bug Hunt BenchSoftware › Software engineering38.727.0−11.7
Dual-Arm Sim (task completion)Tool & computer use › Robotics11.30.0−11.2
FrontierCode 2.0 - ScoreSoftware › Software engineering27.638.9+11.2
FrontierMath Tier 4 (v2) - PrivateMath78.087.8+9.8
Artificial Analysis Intelligence Index v5.0.1Composite indexes50.441.2−9.2
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering74.482.7+8.2
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax30.538.5+8.0
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations36.644.6+8.0
Humanity's Sixth SensePerception & media › Spatial & visual reasoning55.463.4+8.0
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax91.183.2−7.9
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work25.632.8+7.2
BuseyBench SVGPerception & media › Generation7.66.9−7.0
Artificial Analysis Coding Agent Index v2.1Software › Software engineering45.552.3+6.8
HiL-Bench (Human-in-Loop Benchmark)Tool & computer use › General agents62.055.3−6.7
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents10.313.5+6.3
TxBench Antibody Discovery V1 (100 evaluations)Science › Life sciences63.757.7−6.0
EBR-bench v4 (card ban, single agent)Games64.870.5+5.7
SWE-Marathon v1.1Software › Software engineering43.849.4+5.6
LMArena Text Arena EloLanguage & writing › General assistance15211490−5.4
BigFinanceBench private holdout (final-answer accuracy)Professional › Finance, accounting & tax54.749.4−5.3
CritPt v2 (agentic, avg@5)Science › Physical sciences72.978.0+5.1
SimpleQA Verified (Epoch anti-abstention)Knowledge & reasoning › General knowledge76.070.9−5.1
ChartographyPerception & media › Documents & charts71.666.6−5.0
CursorBench 4.1Software › Software engineering56.561.4+4.9
GradeBenchProfessional › Education56.551.9−4.6
DAYJOB: Law (November 2026 release)Professional › Legal5.99.7+3.8
DrugDiscoveryBenchScience › Life sciences65.469.1+3.7
Professional Reasoning Benchmark - LegalProfessional › Legal49.453.1+3.7
SWE-Marathon v1.2Software › Software engineering54.558.1+3.7
Terminal-Bench 4.0Software › Software engineering57.654.0−3.5
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following70.373.7+3.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience › Life sciences64.761.3−3.3
Agents' Last Exam v1.1 (ALE v1.1)Professional › General knowledge work40.243.4+3.3
TxBench Oligo Discovery V1 (113 evaluations)Science › Life sciences51.054.3+3.3
Vals Legal IndexProfessional › Legal76.873.6−3.1
FrontierSWE V2 Best@5Software › Software engineering64.361.2−3.1
CWE-bench v2 Pass@1Software › Cybersecurity65.068.1+3.1
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations77.574.4−3.1
LMArena Code Arena WebDevSoftware › Frontend & design16811703+3.0
OSWorld 2.1 Offline, 500-step Partial ScoreTool & computer use › Computer & browser use74.777.6+2.9
GBENCH One-shot Coding GScoreGames69.766.8−2.8
micro1 PharmaBenchProfessional › General knowledge work8.611.4+2.8
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents23.829.3+2.8
LMArena Agent Arena Steering BurdenTool & computer use › General agents12.217.4+2.6
GDP.xlsxProfessional › General knowledge work38.340.9+2.6
Finance Agent v2.1Professional › Finance, accounting & tax67.765.1−2.5
Professional Reasoning Benchmark - FinanceProfessional › Finance, accounting & tax49.952.3+2.4
FrontierMath Tiers 1-3 (v2) - PrivateMath86.088.1+2.1
MultiChallengeLanguage & writing › Conversation78.780.6+1.9
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents12.615.7+1.5
CursorBench 4.0Software › Software engineering52.954.3+1.4
ARC-AGI-2Knowledge & reasoning › Abstract reasoning91.392.5+1.3
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations50.151.1+1.1
Chess Puzzles (task version 2.0)Games59.060.0+1.0
Mystery Game PuzzlesGames64.065.0+1.0
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax74.875.8+1.0
Legal Research BenchProfessional › Legal54.853.8−1.0
FrontierSWE V2 Worst@5Software › Software engineering44.545.3+0.9
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-6.0-4.5+0.7
LiveBench 2.0 (2026-11-06 release)Composite indexes77.777.0−0.7
AA-Briefcase v1.1 EloProfessional › General knowledge work14851497+0.6
GDP.pdf (November 2026 rescoring)Professional › General knowledge work53.654.2+0.6
Tax Agent BenchProfessional › Finance, accounting & tax79.078.4−0.6
DAYJOB: TaxProfessional › Finance, accounting & tax3.23.6+0.4
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine23.624.0+0.4
GPQA DiamondScience › Multidisciplinary96.496.8+0.4
Epoch Capabilities Index (ECI)Composite indexes167167+0.2
Vals Finance IndexProfessional › Finance, accounting & tax66.066.1+0.1
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge46.146.3+0.1
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.80.70.0
ARC-AGI-1 Semi-PrivateKnowledge & reasoning › Abstract reasoning98.598.50.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
SimpleBenchKnowledge & reasoning › General knowledge91.791.70.0
Data: Artificial Analysis, Surge AI, Epoch AI and 24 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…