Atlas

Compare Models

Claude Fable 5.1 vs Gemini 4 Argon

Gemini 4 Argon leads by 0.2 capability points (≈95% interval −2.3 to +2.7): statistically tied · P(higher) ≈ 56%

Claude Fable 5.1 wins 109 of 209 shared benchmarks (97 lost, 3 tied)

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

Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting.

Claude Fable 5.1vsGemini 4 Argoncapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins109 – 97 · 209 shared, 3 tied

Comparing matching setups only (42 benchmarks excluded) · suites counted once
← Claude Fable 5.1Gemini 4 Argon →

Claude Fable 5.1vsGemini 4 Argonwins by categoryData Nov 15, 2026Data: Vals AI, Artificial Analysis, Center for AI Safety and 63 othersVillage LLM Atlasatlas.village.dev

Margins

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

VCT-Refusal (CAIS)Safety98.910.7−88.2
AA-PersonalityBench Transparency indexChat & Writing98.121.9−76.9
HPCT Refusal (CAIS)Safety100.029.0−71.0
HieroglyphBenchMultimodal20.469.7+49.3
AA-PersonalityBench Slop indexChat & Writing23.867.6+44.2
BioSecBench-Refusal V1 Clean RerunsSafety5.648.7+43.1
AA-PersonalityBench Imagination indexChat & Writing86.746.7−40.4
CheatBenchSafety45.88.6+37.2
AA-PersonalityBench Conviction indexChat & Writing10063.8−36.5
LLM Creative Story WritingChat & Writing3.80.8−33.4
Concept Synth INDUCTION Challenge64 Round 1General QA34.465.6+31.3
ChessBench CoherenceGames61.992.6+30.7
AutomationBench 1.0.6 (Zapier)Agents22.450.1+27.7
BrokenArXiv 08/2026 (Native Harnesses)Math79.854.2−25.6
ChartographyMultimodal46.271.6+25.4
MASKSafety85.961.0−24.9
ChessBenchGames45.370.0+24.7
Agent Red Teaming k=100 (CAIS)Safety34.210.2+24.0
Concept Synth INDUCTION Challenge100 Round 1General QA33.057.0+24.0
Terminal-Bench 2.1Code81.657.7−23.9
ClockBench AccuracyMultimodal59.483.3+23.9
LisanBench Path LengthGames151007560−23.4
ChessBench EloGames10951485+23.2
TextQuests Game ProgressGames56.677.3+20.7
EnigmaEvalGeneral QA44.264.6+20.4
Creative Writing v3 EloChat & Writing21551758−20.1
TaxCalcBench TY25 (web search)Finance & Accounting54.074.0+20.0
Agent Initiative BenchmarkAgents7.55.6−19.0
Finance Agent v2 - PrecedentsFinance & Accounting32.249.8+17.6
DAYJOB: LawLegal26.59.9−16.6
LMArena Agent Arena Bash RecoveryAgents11.6-21.4−16.5
SpatialViz-BenchMultimodal77.994.2+16.3
ERQAMultimodal70.386.5+16.2
MineBench — Bradley-Terry RatingGames19282195+16.2
APEX-Agents 1.1Professional Work71.887.4+15.6
AA-PersonalityBench Warmth indexChat & Writing94.379.1−15.4
GDP.xlsxProfessional Work23.138.3+15.2
MirrorCode (ML, +Private, 2L)Code73.358.9−14.4
APEX-SWECode63.649.2−14.4
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science3.617.6+13.9
AA-PersonalityBench Informativeness indexChat & Writing96.282.9−13.5
AA-PersonalityBench Expressiveness indexChat & Writing98.184.8−13.5
NoCoT-Bench NCRI 15.2General QA128151+12.7
Blueprint-Bench 2Multimodal0.40.5+12.5
BioTIER RefusalSafety99.587.2−12.3
MindCube TinyMultimodal84.596.7+12.2
sudo L7Code31.118.9−12.2
Ramp Accounting BenchFinance & Accounting49.761.1+11.4
ChessBench AccuracyGames71.381.8+11.3
Dual-Arm Sim (task completion)Agents0.011.3+11.3
BullshitBench v2 LeaderboardSafety77.066.0−11.0
Chess Puzzles (task version 2.0)Games48.059.0+11.0
Time Horizon Index: KSPAgents63.352.5−10.8
HANDBOOK.md Agents v1.1Professional Work47.136.6−10.5
RiemannBenchMath65.676.0+10.4
Finance Agent v2 - Market AnalysisFinance & Accounting69.579.7+10.2
SnorkelLegalLegal35.946.0+10.1
CWE-bench v1 Programmatic Pass@1Cybersecurity58.068.0+10.0
Eval Connections One-Shot with TrapsGames82.092.0+10.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal17.427.4+10.0
FrontierMath Tier 4 (v2) - PrivateMath87.878.0−9.8
Finance Agent v2 - Earnings AnalysisFinance & Accounting75.584.8+9.3
DAYJOB: AccountingFinance & Accounting27.218.0−9.2
Humanity's Last ExamGeneral QA46.555.6+9.1
AA-Robotics v0.1Agents46.355.3+9.0
AA-AnalystAgent (pass^5)Professional Work57.548.8−8.7
BioSecBench-Structure V1 (58 evaluations)Science6.314.9+8.6
CUA-benchAgents13.24.8−8.3
SnorkelManufacturingProfessional Work11.319.5+8.2
ComplexConstraintsChat & Writing45.153.3+8.2
HealthBench (length-adjusted, Artificial Analysis)Health & Medicine57.249.1−8.2
Finance Agent v2 - All PassFinance & Accounting47.855.8+8.0
Last Translation Benchmark (LTBv1-eval, blind)Chat & Writing32.940.9+8.0
LMArena Code Arena WebDevCode17421682−7.9
Machiavelli Deception (CAIS)Safety82.485.0−7.8
AA-LCR v1.1Search87.579.7−7.8
LisanBench Difficulty-WeightedGames45733849−7.8
LMArena Agent Arena Bash ReliabilityAgents10.1-5.4−7.7
LMArena Agent Arena Net Improvement (Steering Burden)Agents14.810.7−7.7
GDPval Extended (Mercor held-out rubric score)Professional Work68.776.4+7.7
EBR-bench v4 (card ban, single agent)Games57.164.8+7.6
Immigration BenchLegal75.468.1−7.4
Senior SWE-BenchCode34.727.7−7.0
FrontierCode 2.0 - Pass RateCode37.630.6−7.0
IntPhys 2Multimodal67.774.6+6.9
Visa Agentic Commerce BenchmarkAgents75.682.2+6.7
FrontierCode 2.0 - ScoreCode34.027.6−6.4
DAYJOB: TaxFinance & Accounting9.63.2−6.4
Real-SWE 1.1Code45.639.2−6.4
ACCoRDGeneral QA83.777.3−6.4
Finance Agent v2 - AdjustmentsFinance & Accounting54.460.5+6.1
Mystery Game PuzzlesGames58.064.0+6.0
DualEntry Accounting AI BenchmarkFinance & Accounting85.191.1+6.0
SpeciEval (Pooled English Runs)Safety87.281.3−6.0
WeirdML v3Science26.020.3−5.7
Vals RSI Index v1.1 (four tasks)Agents36.130.6−5.5
Snorkel WorkplaceAgentsProfessional Work15.821.2+5.4
Code2Video-AA EloCode13251259−5.3
PCT Political Inconsistency (CAIS)Safety33.438.7−5.3
BuseyBench SVGMultimodal7.07.6+5.3
MedCodeHealth & Medicine53.558.8+5.3
ArXivMath 08/2026 (Native Harnesses)Math87.782.5−5.3
SimpleQA Verified (Epoch anti-abstention)General QA70.876.0+5.2
SAGEMath48.553.6+5.1
OpenWorldEngineer (Mean@8)Code29.124.1−5.0
AA-Briefcase v1.1 EloAgents15771485−5.0
HardGeoBench <=100miMultimodal53.858.8+5.0
Robo-benchAgents5.80.8−5.0
RWE-benchAgents7.012.0+5.0
HealthBench ProfessionalHealth & Medicine69.264.2−5.0
Finance Agent v2 - ComparablesFinance & Accounting46.051.0+5.0
Build-A-Chip-Bench (20-task index run)Agents23.518.6−4.9
GameBench 2Games94.089.2−4.9
CritPt v2 (agentic, avg@5)Science77.772.9−4.9
APEX-Accounting 1.1Finance & Accounting66.771.5+4.8
Snorkel Digital Work IndexIndexes23.628.3+4.7
SnorkelFinance 2.0Finance & Accounting19.615.1−4.5
Finance Agent v2 - Financial ModelingFinance & Accounting25.329.8+4.5
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine18.313.9−4.4
SnorkelRevOpsProfessional Work14.018.4+4.4
Longform Creative WritingChat & Writing85.380.9−4.4
Finance Agent v2 - Disclosure AnalysisFinance & Accounting70.374.5+4.2
FrontierMath Tiers 1-3 (v2) - PrivateMath90.286.0−4.2
DeepFinanceFinance & Accounting45.641.4−4.2
micro1 PharmaBenchHealth & Medicine12.88.6−4.2
τ³-Banking (Mercor run, bm25_grep retrieval)Agents54.058.1+4.1
SWE Atlas v1.1 - OverallCode74.170.1−4.0
CritPt (avg@5)Science31.127.1−4.0
LiveBench 2.0Agents88.084.1−3.9
MedScribeHealth & Medicine91.387.4−3.9
AA-PersonalityBench Organization indexChat & Writing60.056.2−3.8
HealthBenchHealth & Medicine63.960.0−3.8
HackerRank-BenchCode45.949.7+3.8
LMArena Agent Arena Praise vs ComplaintAgents31.624.1−3.8
ALE-BenchCode28492747−3.6
Creative Writing v3 Rubric ScoreChat & Writing16.315.6−3.5
MCP Atlas April 2026 (all 1,000 tasks)Agents87.290.7+3.5
Conceptual Reasoning IndexIndexes74.972.4−3.5
LMArena Text Arena EloChat & Writing15031522+3.4
TextQuests Average Cumulative HarmSafety27.126.1+3.3
TxBench-Antibody-Development V1 (60 evaluations)Science27.831.1+3.3
HardGeoBench Avg DistanceMultimodal845741+3.2
DAXBench v2Code91.588.3−3.2
FrontierCode 1.1 Extended - ScoreCode63.660.6−3.0
CloudOpsBench v1.0Agents60.657.5−3.0
FrontierPhysics v0.1Science11.38.3−3.0
DistressBenchSafety71.568.6−2.9
DAYJOB: Data ScienceProfessional Work26.023.3−2.7
FrontierCode 1.1 Extended - Pass RateCode68.866.1−2.7
Financial Coaching BenchFinance & Accounting75.773.1−2.6
Westworld Finance Diligence BenchFinance & Accounting58.260.7+2.5
Harvey LAB-AA v1.1 (hallucination-gated)Legal6.43.9−2.5
Furniture Assembly BenchmarkMultimodal70.072.5+2.5
FrontierML ($200 wallet)Agents41.939.5−2.4
MysteryMechanismScience47.745.5−2.3
MMMU ProMultimodal86.188.3+2.2
Bug Hunt BenchCode41.038.7−2.2
Snorkel Agentic Coding 2.0Code39.641.8+2.2
Professional Reasoning Benchmark - LegalLegal51.649.4−2.2
Toloka Arena (v3)Agents71.073.0+2.0
Humanity's Last Exam DiamondGeneral QA51.353.2+1.9
Excel Modeling Benchmark v1.1Finance & Accounting76.774.8−1.9
SWE-Marathon v1.1Code45.643.8−1.9
Snorkel SWE-bench CLICode14.516.3+1.8
HoneyBench v0.2 (hack rate)Safety61.863.6−1.8
LMCAGeneral QA65.563.7−1.8
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting79.781.5+1.8
AA-LCR Extended (Mercor held-out)Search73.074.7+1.7
FrontierMath ErdősMath0.01.5+1.5
GBENCH One-shot Coding GScoreGames69.367.9−1.4
Excel Modeling BenchmarkFinance & Accounting76.775.2−1.4
RSI-Exam 0.1 (Full 88)Agents0.50.5+1.3
SciCode 1.0.1 (Artificial Analysis)Code63.161.8−1.3
ARC-AGI-2General QA90.091.3+1.3
HardGeoBench Country AccuracyMultimodal61.362.5+1.2
DAYJOB: HealthcareHealth & Medicine9.610.8+1.2
GDPxProfessional Work8.57.3−1.2
Tax Agent BenchFinance & Accounting77.979.0+1.2
AA-3D v1.0 EloCode11391147+1.1
CursorBench 4.0Code51.852.9+1.1
ARC-AGI-1 Semi-PrivateGeneral QA97.598.5+1.0
DAYJOB: FinanceFinance & Accounting19.818.8−1.0
ProofBench v1.1Math100.099.0−1.0
WellnessBenchSafety51.050.0−1.0
Professional Reasoning Benchmark - FinanceFinance & Accounting50.849.9−0.9
Remote Labor Index (RLI)Professional Work17.917.1−0.8
Design Arena Overall Code CategoriesCode13311326−0.8
LMArena Agent Arena Steering BurdenAgents14.813.2−0.8
AutoResearchExam (hidden-test AUARC)Agents0.60.6−0.8
MissionReportProfessional Work63.364.1+0.8
MedXpertQA MM Extended (Mercor held-out)Health & Medicine66.066.7+0.7
SnorkelUnderwrite 2.0Finance & Accounting24.023.3−0.7
GDPval (Mercor 213-task rubric score)Professional Work86.685.9−0.7
Arena Alignment IndexSafety80.179.4−0.7
LMArena Agent Arena Confirmed SuccessAgents14.413.2−0.6
HiL-Bench (Human-in-Loop Benchmark)Agents61.562.0+0.5
Terminal-Bench 4.0Code57.957.4−0.5
GDPval-AA v2.1Professional Work16441634−0.5
Epoch Capabilities Index (ECI)Indexes167167−0.4
Prophet Arena Arena Score (Fixed Context)Forecasting85.085.3+0.3
LegalBenchLegal88.588.3−0.2
LegalReasonBenchLegal50.650.8+0.2
LMArena Agent Arena Tool HallucinationAgents0.80.5−0.1
NoCoT-Bench NCKI (kspine v3)General QA1281280.0
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting86.086.00.0
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting77.077.00.0
DTBenchGeneral QA97.697.60.0
HardGeoBench <=10miMultimodal45.045.00.0
OTIS Mock AIME 2024-2025Math100.0100.00.0

Claude Fable 5.1vsGemini 4 Argonnormalised score marginsData Nov 15, 2026Data: Vals AI, Artificial Analysis, Center for AI Safety and 63 othersVillage LLM Atlasatlas.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…