Atlas

Compare Models

Claude Fable 5.1 vs GPT-5.6 Sol

Claude Fable 5.1 leads by 2.9 capability points (≈95% interval +0.5 to +5.2): a clear lead over GPT-5.6 Sol · P(higher) ≈ 99%

Wins 165 of 216 shared benchmarks (45 lost, 6 tied)

Capability vs. relative cost
Claude Fable 5.1GPT-5.6 SolClosed 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 5.1vsGPT-5.6 Solcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins165 – 45 · 216 shared, 6 tied

Comparing matching setups only (102 benchmarks excluded) · suites counted once
← Claude Fable 5.1GPT-5.6 Sol →

Claude Fable 5.1vsGPT-5.6 Solwins by categoryData Nov 15, 2026Data: Vals AI, Center for AI Safety, Arena and 72 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-5.6 Sol minus Claude Fable 5.1.

VCT-Refusal (CAIS)Safety98.911.3−87.6
HPCT Refusal (CAIS)Safety100.024.0−76.0
KernelBench Mega Normalized (RTX PRO 6000)Code64.77.4−57.3
MirrorCode (ML, +Private, 2L)Code73.320.0−53.3
TasteVal (compute multiplier)Science1.70.6−49.4
Medical Long Context Reasoning (MLCR-AA)Health & Medicine71.126.1−45.0
Olam Social Poker EloGames14941534+43.5
Time Horizon Index: KSPAgents63.323.8−39.5
SciUniverse Level 1Science45.39.4−35.9
LisanBench Difficulty-WeightedGames45731403−34.0
LisanBench Path LengthGames151004699−32.3
EnterpriseBench: CoreCraftProfessional Work77.446.2−31.2
Document ArenaProfessional Work15151480−31.1
Terminal-Bench-Science 0.1Science52.622.4−30.2
HiL-Bench (Human-in-Loop Benchmark)Agents61.532.3−29.2
CheatBenchSafety45.874.9−29.1
BullshitBench v2 LeaderboardSafety77.048.0−29.0
SM BenchSafety89.061.1−27.9
BioTIER RefusalSafety99.571.9−27.6
ChessBench CoherenceGames61.989.1+27.2
Gray Swan IPI Attack Success (k=15)Safety1.027.0−26.0
LMArena Agent Arena Net ImprovementAgents14.16.3−25.7
Concept Synth INDUCTION Challenge64 Round 1General QA34.457.8+23.4
Terminal-Bench Science 0.1 (Vals)Science34.312.9−21.4
Terminal-Bench 4.0Code57.937.3−20.6
FrontierCUA pass@1Agents36.716.4−20.3
Humanity's Last Exam DiamondGeneral QA51.331.2−20.1
BrokenArXiv 06/2026Math86.667.3−19.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents14.84.6−19.2
BabyVisionMultimodal46.465.5+19.1
Terminal-Bench 2.1 Extended (Mercor held-out)Code19.537.4+17.9
FrontierSWE V2 (Mercor rerun, mean@3)Code37.220.2−17.0
ProofBench v1.1Math100.083.0−17.0
Search ArenaSearch12371258+16.5
Humanity's Last ExamGeneral QA60.944.5−16.4
APEX-SWECode63.648.0−15.6
ALE-BenchCode28492410−15.6
OSWorld 2.0 - reported partial reward, patch and step limit unspecifiedAgents77.962.6−15.3
HANDBOOK.mdProfessional Work38.823.5−15.3
HardGeoBench Avg DistanceMultimodal8881381−15.2
Drone-BenchAgents91.276.1−15.1
HardGeoBench <=10miMultimodal45.030.0−15.0
LLM Creative Story WritingChat & Writing3.82.5−14.6
TextQuests Average Cumulative HarmSafety27.131.4−14.5
MysteryMechanismScience47.733.3−14.4
BioSecBench-Structure V1 (58 evaluations)Science6.320.7+14.4
DAYJOB: FinanceFinance & Accounting19.85.5−14.3
GameBench 2Games94.080.1−14.0
HardGeoBench Country AccuracyMultimodal61.347.5−13.8
Antidote: Everyday EditionSafety10651027−13.6
LABBench2Science68.682.1+13.5
Furniture Assembly BenchmarkMultimodal70.056.7−13.3
LLM Debate BenchmarkChat & Writing17381645−13.2
Toloka Arena (v1)Agents60.747.6−13.1
LMArena Agent Arena Praise vs ComplaintAgents31.66.3−12.6
HardGeoBench <=100miMultimodal53.841.3−12.5
EBR-bench v4 (card ban, single agent)Games57.144.8−12.4
BrowseComp (Mercor 130-question subset)Search78.390.6+12.3
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting77.064.9−12.0
CadQueryEval (Corrected Geometry)Code76.088.0+12.0
TaxCalcBench TY25Finance & Accounting42.030.0−12.0
SpeciEval (Refusal-Aware Retries, Historical)Safety87.299.0+11.8
SpeciEval (Pooled English Runs)Safety87.299.0+11.8
RT-SAFE Collisions per EpisodeSafety20.925.6−11.8
GSO Opt@1Code88.276.5−11.8
LiveBench 2.0Agents88.076.3−11.7
SWE-TogetherCode69.357.8−11.5
Chess Puzzles (task version 2.0)Games48.059.0+11.0
AA-Briefcase EloAgents16621475−10.9
InteractiveBench (task pass)Agents41.730.8−10.8
Hemingway-benchChat & Writing10951057−10.5
ProgramBench (Mercor 100-task subset)Code11.31.0−10.3
NoCoT-Bench NCRI 15.2General QA128109−10.2
SWE Atlas - Codebase QnACode43.353.5+10.2
CursorBench 4.0Code51.841.7−10.1
Concept Synth INDUCTION Challenge100 Round 1General QA33.043.0+10.0
Toloka Arena (v3)Agents71.061.0−10.0
AA-AnalystAgent (pass^5)Professional Work57.547.5−10.0
Tax Agent BenchFinance & Accounting77.968.0−9.9
MedCodeHealth & Medicine53.544.0−9.5
APEX-AccountingFinance & Accounting61.051.5−9.5
MASKSafety85.995.4+9.5
BuseyBench SVGMultimodal7.06.1−9.2
BenchCAD - geometric overlapMultimodal92.683.4−9.2
GeoBenchMultimodal43033843−9.2
Creative Writing v3 EloChat & Writing21551974−9.2
HackerRank-BenchCode47.538.4−9.1
FrontierFinance v3.1Finance & Accounting55.946.8−9.0
Shroombench (mushroom identification, top-1)Multimodal51.042.0−9.0
AutoResearchExam (hidden-test AUARC)Agents0.60.5−8.9
ClockBench AccuracyMultimodal59.468.3+8.9
TxBench-Antibody-Development V1 (60 evaluations)Science27.818.9−8.9
RiemannBenchMath65.674.4+8.8
HealthBenchHealth & Medicine63.955.3−8.6
Conceptual Reasoning IndexIndexes74.968.7−8.6
Design Creativity BenchCode48.651.3+8.4
MMLU Pro - LawGeneral QA87.579.1−8.4
PCT Political Inconsistency (CAIS)Safety33.441.7−8.3
Blueprint-Bench 2Multimodal0.40.3−8.3
ActiveVisionMultimodal15.323.5+8.2
DAYJOB: HealthcareHealth & Medicine9.61.6−8.0
LibraryDesignBenchCode47.539.5−8.0
SciCode (main problems)Science24.216.2−8.0
Terminal-Bench 2.1Code81.689.5+7.9
LiveCodeBench (Vals AI public dataset)Code90.582.6−7.9
GDPval-AA v2Professional Work17451585−7.9
ACCoRDGeneral QA83.775.9−7.8
MindCube TinyMultimodal84.576.8−7.7
ATLAS-Finance v1.0Finance & Accounting12.04.7−7.3
ChessBench EloGames10951218+7.3
CVDP (517-problem set)Code62.655.4−7.2
HieroglyphBenchMultimodal20.427.6+7.2
LMCAGeneral QA65.558.5−7.0
Finance Agent v2 - Disclosure AnalysisFinance & Accounting70.363.3−6.9
ITBench-AAAgents49.556.2+6.7
GDPval-AA v2.1Professional Work17651619−6.4
FORTRESS Average Risk Score (ARS)Safety13.720.1−6.4
Finance Agent v2 - All PassFinance & Accounting47.841.4−6.4
AutomationBench 1.0.6 (Zapier)Agents22.428.8+6.4
ATLAS-Finance v1.1Finance & Accounting10.34.0−6.3
Artificial Analysis Intelligence Index v4.3Indexes53.447.1−6.3
SWE-Clar (clarification)Code84.378.1−6.2
Agent Red Teaming k=100 (CAIS)Safety34.240.3−6.1
LMArena Agent Arena Confirmed SuccessAgents14.42.3−6.1
MedScribeHealth & Medicine91.385.2−6.1
Finance Agent v2 - PrecedentsFinance & Accounting32.226.2−6.0
WellnessBenchSafety51.057.0+6.0
OpenDS-WorldSearch33.528.0−5.5
ChessBench AccuracyGames71.376.4+5.5
LMArena Agent Arena Bash RecoveryAgents11.60.7−5.4
DeepSWE v1.1Code67.372.7+5.4
MCP Atlas April 2026 (all 1,000 tasks)Agents87.281.8−5.4
ComplexConstraintsChat & Writing45.150.5+5.4
Finance Agent v2 - Market AnalysisFinance & Accounting69.564.2−5.4
NoCoT-Bench NCKI (kspine v3)General QA128121−5.4
SciCode 1.0.1 (Artificial Analysis)Code63.157.8−5.3
Finance Agent v2 - AdjustmentsFinance & Accounting54.449.2−5.2
FrontierMath Tier 4 (v2) - PrivateMath87.882.9−4.9
Hedge-BenchFinance & Accounting26.831.5+4.7
LMArena Agent Arena Steering BurdenAgents14.85.5−4.6
Finance Agent v2 - Financial ModelingFinance & Accounting25.320.7−4.6
TextQuests Game ProgressGames56.652.1−4.5
MMLU Pro - PsychologyGeneral QA93.789.2−4.5
MMLU Pro - EconomicsGeneral QA95.190.6−4.5
Excel Modeling BenchmarkFinance & Accounting76.772.3−4.3
CritPt v2 (agentic, avg@5)Science72.977.1+4.3
MMLU Pro - Computer ScienceGeneral QA94.990.7−4.1
WeirdML (v2)Science92.988.8−4.1
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting79.775.6−4.1
SAGEMath48.552.6+4.0
Arena Alignment IndexSafety80.184.2+4.0
MMLU Pro - PhysicsGeneral QA96.792.8−3.9
Extended NYT Connections 940 (Quadratic v1)Games90.093.8+3.8
IntPhys 2Multimodal67.763.9−3.8
SUPERChemScience74.771.0−3.7
Longform Creative WritingChat & Writing85.381.7−3.6
DeepResearch Bench II (Mercor 130-task set)Search55.051.5−3.5
AA-LCR v1.1Search87.584.0−3.5
MortgageTax (Vals all 1,258 documents)Finance & Accounting70.867.3−3.5
MMMU ProMultimodal86.182.7−3.4
MMLU Pro - HistoryGeneral QA85.381.9−3.4
Realm Financial ReasoningFinance & Accounting47.844.4−3.4
MMLU Pro (Vals 5-Shot CoT)General QA92.489.1−3.3
Epoch Capabilities Index (ECI)Indexes167163−3.2
SAB ResearchFinance & Accounting96.893.6−3.2
LMArena Text Arena EloChat & Writing15031485−3.1
SWE-Clar (implementation)Code60.057.0−3.0
FrontierCode 1.1 Extended - ScoreCode63.660.6−3.0
MMLU Pro - EngineeringGeneral QA90.887.8−3.0
SpatialViz-BenchMultimodal77.980.7+2.8
ObviousBenchGeneral QA99.396.5−2.8
CritPt (avg@5)Science31.128.6−2.6
MMLU Pro - HealthGeneral QA86.784.1−2.6
ReactBench (39 tasks)Code49.246.7−2.6
ARC-AGI-2General QA90.092.5+2.5
ArXivMath 06/2026Math91.088.5−2.4
Roboflow Vision Evals (October 2026, five tasks)Multimodal74.772.4−2.3
Roboflow Vision Evals (September 2026 protocol)Multimodal81.379.0−2.3
Clocktower BenchGames15671583+2.3
MMLU Pro - OtherGeneral QA89.987.7−2.3
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science3.65.8+2.1
MMLU Pro - MathMath97.395.3−2.1
GlobeBenchGeneral QA21.623.7+2.0
DistressBenchSafety71.573.6+2.0
GPQA DiamondGeneral QA92.994.9+2.0
GeneBench-Pro Public Release SubsetScience26.024.0−2.0
MMLU Pro - BusinessGeneral QA95.293.3−1.9
DTBenchGeneral QA97.695.7−1.9
MMLU Pro - ChemistryGeneral QA93.892.0−1.9
MMLU Pro - PhilosophyGeneral QA90.288.4−1.8
GBENCH One-shot Coding GScoreGames69.367.7−1.6
LegalBenchLegal88.587.0−1.5
MMLU Pro - BiologyGeneral QA96.194.6−1.5
ChartographyMultimodal46.245.0−1.2
TaxEval v2Finance & Accounting76.074.8−1.2
Finance Agent v2 - Earnings AnalysisFinance & Accounting75.574.4−1.1
FrontierMath Tiers 1-3 (v2) - PrivateMath90.289.1−1.1
SAB RankingForecasting-0.1-0.1−1.0
Eval Connections One-Shot with TrapsGames82.083.0+1.0
LMArena Agent Arena SteerabilityAgents8.46.5−0.9
Machiavelli Deception (CAIS)Safety82.482.7−0.9
ERQAMultimodal70.371.2+0.9
Financial Coaching BenchFinance & Accounting75.774.9−0.8
Ruby LLM Benchmarks — Program FixerCode76.075.2−0.8
Finance Agent v2 - ComparablesFinance & Accounting46.045.3−0.7
LMArena Agent Arena Bash ReliabilityAgents10.111.4+0.7
Creative Writing v3 Rubric ScoreChat & Writing16.316.2−0.5
Vision Arena OverallMultimodal12891288−0.3
Prophet Arena Arena Score (Fixed Context)Forecasting85.084.8−0.2
LMArena Agent Arena Tool HallucinationAgents0.81.0+0.1
ARC-AGI-1 Semi-PrivateGeneral QA97.597.50.0
BabaIsBench Stage 0: The IntroGames100.0100.00.0
FrontierMath ErdősMath0.00.00.0
Mystery Game PuzzlesGames58.058.00.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
Senior SWE-BenchCode34.734.70.0

Claude Fable 5.1vsGPT-5.6 Solnormalised score marginsData Nov 15, 2026Data: Vals AI, Center for AI Safety, Arena and 72 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…