Atlas

Compare Models

Claude Fable 5.1 vs Gemini 4 Hydrogen

Gemini 4 Hydrogen leads by 0.1 capability points (≈95% interval −2.4 to +2.7): statistically tied · P(higher) ≈ 54%

Wins 31 of 58 shared benchmarks (24 lost, 3 tied)

Capability vs. relative cost
Claude Fable 5.1Gemini 4 HydrogenClosed weightsOpen weights

Benchmark wins24 – 31 · 58 shared, 3 tied

Comparing matching setups only (5 benchmarks excluded) · suites counted once
← Claude Fable 5.1Gemini 4 Hydrogen →
Data: Surge AI, Epoch AI, Arena and 21 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Gemini 4 Hydrogen minus Claude Fable 5.1.

TxBench-Antibody-Development V1 (60 evaluations)Science › Life sciences27.860.6+32.8
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations22.451.1+28.8
ChartographyPerception & media › Documents & charts46.266.6+20.4
GDP.xlsxProfessional › General knowledge work23.140.9+17.8
EBR-bench v4 (card ban, single agent)Games57.170.5+13.3
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following60.473.7+13.3
DAYJOB: Law (November 2026 release)Professional › Legal22.59.7−12.8
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning70.057.5−12.5
Chess Puzzles (task version 2.0)Games48.060.0+12.0
Bug Hunt BenchSoftware › Software engineering41.031.1−9.8
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine18.327.2+8.9
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine31.224.0−7.2
Mystery Game PuzzlesGames58.065.0+7.0
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering63.870.8+7.0
LiveBench 2.0 (2026-11-06 release)Composite indexes84.077.0−7.0
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents9.3-4.5−6.9
DAYJOB: AccountingProfessional › Finance, accounting & tax27.233.6+6.4
HiL-Bench (Human-in-Loop Benchmark)Tool & computer use › General agents61.555.3−6.2
DAYJOB: TaxProfessional › Finance, accounting & tax9.63.6−6.0
SpireClimbGames2.07.7+5.7
FrontierCode 2.0 - Pass RateSoftware › Software engineering37.643.2+5.6
LMArena Code Arena WebDevSoftware › Frontend & design17411703−5.1
SimpleBenchKnowledge & reasoning › General knowledge86.691.7+5.1
FrontierCode 2.0 - ScoreSoftware › Software engineering34.038.9+4.8
GDP.pdf (November 2026 rescoring)Professional › General knowledge work49.454.2+4.8
AA-Briefcase v1.1 EloProfessional › General knowledge work15771497−4.4
CritPt v2 (agentic, avg@5)Science › Physical sciences76.680.6+4.0
SWE-Marathon v1.1Software › Software engineering45.649.4+3.8
FORTRESS Average Risk Score (ARS)Safety & alignment13.717.5−3.7
ARC-AGI-2Knowledge & reasoning › Abstract reasoning90.092.5+2.5
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations47.144.6−2.5
CursorBench 4.0Software › Software engineering51.854.3+2.5
LMArena Text Arena EloLanguage & writing › General assistance15021490−2.2
FrontierMath Tiers 1-3 (v2) - PrivateMath90.288.1−2.1
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax36.538.5+2.0
TaxCalcBench TY25Professional › Finance, accounting & tax42.044.0+2.0
LMArena Agent Arena Steering BurdenTool & computer use › General agents13.517.4+2.0
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax85.183.2−1.9
BuseyBench SVGPerception & media › Generation7.06.9−1.7
Professional Reasoning Benchmark - FinanceProfessional › Finance, accounting & tax50.852.3+1.5
Professional Reasoning Benchmark - LegalProfessional › Legal51.653.1+1.5
micro1 PharmaBenchProfessional › General knowledge work12.811.4−1.4
GradeBenchProfessional › Education53.151.9−1.2
ARC-AGI-1 Semi-PrivateKnowledge & reasoning › Abstract reasoning97.598.5+1.0
Gray Swan IPI Attack Success (k=15)Safety & alignment1.00.0+1.0
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.775.8−0.9
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents14.215.7+0.8
Tax Agent BenchProfessional › Finance, accounting & tax77.978.4+0.6
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents28.129.3+0.6
sudo L7 (October 2026 rescoring)Software › Software engineering32.231.7−0.5
GBENCH One-shot Coding GScoreGames67.366.8−0.5
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents13.713.5−0.3
Epoch Capabilities Index (ECI)Composite indexes167167−0.2
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.00.7−0.1
SimpleQA Verified (Epoch anti-abstention)Knowledge & reasoning › General knowledge70.870.9+0.1
Dual-Arm Sim (task completion)Tool & computer use › Robotics0.00.00.0
FrontierMath Tier 4 (v2) - PrivateMath87.887.80.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
Data: Surge AI, Epoch AI, Arena and 21 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…