Atlas

Compare Models

Gemini 4 Hydrogen vs GPT-6 Astra

GPT-6 Astra leads by 1.4 capability points (≈95% interval −1.2 to +4.1): statistically tied · P(higher) ≈ 85%

Wins 38 of 76 shared benchmarks (36 lost, 2 tied)

Capability vs. relative cost
Gemini 4 HydrogenGPT-6 AstraClosed weightsOpen weights

Benchmark wins36 – 38 · 76 shared, 2 tied

Comparing matching setups only (35 benchmarks excluded) · suites counted once
← Gemini 4 HydrogenGPT-6 Astra →
Data: Artificial Analysis, Epoch AI, Surge AI and 19 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6 Astra minus Gemini 4 Hydrogen.

TxBench-Antibody-Development V1 (60 evaluations)Science › Life sciences60.632.8−27.8
DAYJOB: Law (November 2026 release)Professional › Legal9.736.6+26.9
SpireClimbGames7.734.0+26.3
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning57.580.0+22.5
Mystery Game PuzzlesGames65.084.0+19.0
GDP.xlsxProfessional › General knowledge work40.922.9−18.0
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering70.854.6−16.2
Artificial Analysis Intelligence Index v5.0.1Composite indexes41.256.9+15.6
Tax Agent BenchProfessional › Finance, accounting & tax78.463.3−15.1
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine27.241.7+14.4
Legal Research BenchProfessional › Legal53.839.4−14.4
Chess Puzzles (task version 2.0)Games60.074.0+14.0
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations44.631.4−13.2
FrontierSWE V2 Best@5Software › Software engineering61.273.0+11.8
GBENCH One-shot Coding GScoreGames66.878.4+11.5
Vals Finance IndexProfessional › Finance, accounting & tax66.155.6−10.5
HiL-Bench (Human-in-Loop Benchmark)Tool & computer use › General agents55.365.3+10.0
FrontierMath Tier 4 (v2) - PrivateMath87.897.6+9.8
FrontierSWE V2 Worst@5Software › Software engineering45.355.1+9.8
Humanity's Sixth SensePerception & media › Spatial & visual reasoning63.453.6−9.8
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations51.141.4−9.7
Gray Swan IPI Attack Success (k=15)Safety & alignment0.08.5−8.5
SimpleBenchKnowledge & reasoning › General knowledge91.783.6−8.1
LiveBench 2.0 (2026-11-06 release)Composite indexes77.084.3+7.3
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering53.559.6+6.1
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations74.468.5−5.9
EBR-bench v4 (card ban, single agent)Games70.576.2+5.7
FrontierMath Tiers 1-3 (v2) - PrivateMath88.193.7+5.6
Terminal-Bench 4.0Software › Software engineering54.059.6+5.6
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents-4.56.5+5.5
FORTRESS Average Risk Score (ARS)Safety & alignment17.512.1+5.3
Vals Legal IndexProfessional › Legal73.668.4−5.2
Terminal-Bench-Science 0.1Science › Multidisciplinary49.554.8+5.2
BigFinanceBench private holdout (final-answer accuracy)Professional › Finance, accounting & tax49.454.6+5.2
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge58.863.8+5.0
Professional Reasoning Benchmark - FinanceProfessional › Finance, accounting & tax52.347.5−4.8
Professional Reasoning Benchmark - LegalProfessional › Legal53.148.4−4.7
micro1 PharmaBenchProfessional › General knowledge work11.416.1+4.7
SimpleQA Verified (Epoch anti-abstention)Knowledge & reasoning › General knowledge70.975.6+4.7
Finance Agent v2.1Professional › Finance, accounting & tax65.160.4−4.7
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering82.778.0−4.6
BuseyBench SVGPerception & media › Generation6.97.3+4.6
ChartographyPerception & media › Documents & charts66.671.0+4.4
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine24.028.4+4.4
HLAB-Vals-1.1Professional › Legal11.37.1−4.2
GDP.pdf (November 2026 rescoring)Professional › General knowledge work54.250.4−3.8
DAYJOB: AccountingProfessional › Finance, accounting & tax33.630.0−3.6
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax75.872.7−3.1
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following73.770.8−2.9
CritPt v2 (agentic, avg@5)Science › Physical sciences80.683.4+2.9
LMArena Text Arena EloLanguage & writing › General assistance14901475−2.6
ARC-AGI-2Knowledge & reasoning › Abstract reasoning92.595.0+2.5
Dual-Arm Sim (task completion)Tool & computer use › Robotics0.02.5+2.5
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering39.041.4+2.4
TxBench Oligo Discovery V1 (113 evaluations)Science › Life sciences54.351.9−2.4
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents13.514.6+2.0
TxBench Preclinical Pharmacology V1 Clean RerunsScience › Life sciences61.359.3−2.0
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax83.281.2−2.0
TaxCalcBench TY25Professional › Finance, accounting & tax44.046.0+2.0
Epoch Capabilities Index (ECI)Composite indexes167169+1.9
LMArena Agent Arena Steering BurdenTool & computer use › General agents17.413.9−1.7
GradeBenchProfessional › Education51.953.6+1.7
TxBench Antibody Discovery V1 (100 evaluations)Science › Life sciences57.756.0−1.7
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents29.332.3+1.5
AA-Briefcase v1.1 EloProfessional › General knowledge work15691544−1.4
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax38.537.3−1.2
DAYJOB: TaxProfessional › Finance, accounting & tax3.64.8+1.2
GPQA DiamondScience › Multidisciplinary96.895.8−1.0
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge46.344.6−0.8
sudo L7 (October 2026 rescoring)Software › Software engineering31.731.1−0.6
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work32.832.2−0.6
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents15.714.9−0.4
DrugDiscoveryBenchScience › Life sciences69.168.7−0.4
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.71.2+0.2
ARC-AGI-1 Semi-PrivateKnowledge & reasoning › Abstract reasoning98.598.50.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
Data: Artificial Analysis, Epoch AI, Surge AI and 19 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…