Atlas

Compare Models

Claude Fable 5.1 vs Claude Opus 5.5

Claude Opus 5.5 leads by 2.0 capability points (≈95% interval −0.6 to +4.6): statistically tied · P(higher) ≈ 93%

Wins 205 of 285 shared benchmarks (69 lost, 11 tied)

Capability vs. relative cost
Claude Fable 5.1Claude Opus 5.5Closed 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 5.1vsClaude Opus 5.5capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins69 – 205 · 285 shared, 11 tied

Comparing matching setups only (4 benchmarks excluded) · suites counted once
← Claude Fable 5.1Claude Opus 5.5 →

Claude Fable 5.1vsClaude Opus 5.5wins by categoryData Nov 15, 2026Data: Mercor, Artificial Analysis, Center for AI Safety and 93 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Claude Opus 5.5 minus Claude Fable 5.1.

Mercor enterprise risk (implicit-risk unsafe rate)Safety77.822.2+55.6
Concept Synth INDUCTION Challenge64 Round 1General QA34.471.9+37.5
KernelBench Mega Normalized (RTX PRO 6000)Code64.7100.0+35.3
CheatBenchSafety45.811.2+34.5
InferenceBenchCode9.312.1+33.1
Concept Synth INDUCTION Challenge100 Round 1General QA33.065.0+32.0
AA-PersonalityBench Conviction indexChat & Writing10069.5−30.8
HoneyBench v0.2 (hack rate)Safety61.831.6+30.2
Terminal-Bench 2.1 Extended (Mercor held-out)Code19.547.5+28.0
Time Horizon Index: KSPAgents63.391.3+28.0
LisanBench Path LengthGames1510024029+27.7
ActiveVisionMultimodal15.341.2+25.9
TasteVal (compute multiplier)Science1.72.3+25.8
BioSecBench-Structure V1 (58 evaluations)Science5.831.0+25.3
KernelBench CUDA Normalized (RTX PRO 6000)Code51.676.3+24.6
CadQueryEval (Corrected Geometry)Code76.0100.0+24.0
RWE-benchAgents7.029.0+22.0
BioSecBench-RespiratoryVirus-Surveillance V1 (110 evaluations)Science3.625.4+21.8
Web Clone BenchCode13.034.2+21.2
ChartographyMultimodal46.266.3+20.1
ChessBenchGames45.365.3+20.0
AA-PersonalityBench Organization indexChat & Writing60.079.1+19.2
Search ArenaSearch12371213−19.2
Chess Puzzles (task version 2.0)Games48.067.0+19.0
Anthropic ECI (AECI), November 2026 fitIndexes173175+18.9
ChessBench CoherenceGames61.980.1+18.2
ClockBench AccuracyMultimodal59.477.2+17.8
Hedge-BenchFinance & Accounting26.844.2+17.4
Code2Video-AA EloCode13251536+17.1
BrokenArXiv 05/2026Math78.095.0+17.0
HieroglyphBenchMultimodal20.437.4+17.0
FrontierML ($200 wallet)Agents41.958.8+16.9
FrontierPhysics v0.1Science11.327.4+16.1
FrontierCode 2.0 - Pass RateCode36.451.8+15.5
ChessBench EloGames10951349+15.1
DAYJOB: HealthcareHealth & Medicine9.624.7+15.1
Machiavelli Deception (CAIS)Safety82.477.5+14.9
LisanBench Difficulty-WeightedGames45735963+14.9
Android Bench 2.0Code22.737.3+14.6
FrontierCode 2.0 - ScoreCode33.047.3+14.4
EBR-bench v4 (card ban, single agent)Games57.171.4+14.3
Terminal-Bench Science 0.1 (Vals)Science34.348.6+14.3
BullshitBench v2 LeaderboardSafety77.063.0−14.0
EnigmaEvalGeneral QA44.257.8+13.6
BrokenArXiv 06/2026Math86.6100.0+13.4
Visa Agentic Commerce BenchmarkAgents75.662.2−13.3
Furniture Assembly BenchmarkMultimodal70.083.3+13.3
Olam Social Poker EloGames14941506+13.0
Mystery Game PuzzlesGames58.071.0+13.0
MineBench — Bradley-Terry RatingGames19282134+12.5
OpenWorldEngineer (Mean@8)Code29.141.6+12.5
SpatialViz-BenchMultimodal77.989.7+11.8
DAYJOB: Data ScienceProfessional Work26.037.3+11.3
ITBench-AAAgents49.538.2−11.3
Real-SWE 1.1Code45.656.8+11.2
AutomationBench 1.0.6 (Zapier)Agents31.442.5+11.1
sudo L7Code31.141.7+10.6
OSWorld 2.1 offline - reported partial reward, step limit unspecifiedAgents76.887.2+10.4
SPU (Science Prediction Universe)Science-0.5-0.3+10.4
Clipping Bench v1.0Games10682018+10.2
BrowseComp (Mercor 130-question subset)Search78.388.5+10.2
TextQuests Average Cumulative HarmSafety27.124.2+9.9
Agent Red Teaming k=100 (CAIS)Safety34.224.4+9.8
AA-PersonalityBench Warmth indexChat & Writing94.384.8−9.6
LMArena Code Arena WebDevCode17421814+9.5
Blueprint-Bench 2Multimodal0.40.5+9.3
DeepFinanceFinance & Accounting45.654.8+9.2
AA-3D v1.0 EloCode11391202+9.2
AutomationBench-AA - release-reported strict pass rateAgents32.141.2+9.1
PostTrainBench Average (v1.1)Indexes40.249.3+9.1
CWE-bench v1 Programmatic Pass@1Cybersecurity58.067.0+9.0
Artificial Analysis Intelligence Index v5.0Indexes49.058.0+9.0
LMArena Agent Arena Net Improvement (Steering Burden)Agents14.819.5+8.9
Design Arena Overall Code CategoriesCode13311384+8.8
Robo-benchAgents5.814.6+8.8
ProgramBench (Mercor 100-task subset)Code11.320.0+8.7
MindCube TinyMultimodal84.593.1+8.6
ChessBench AccuracyGames71.379.3+8.6
VCT-Refusal (CAIS)Safety98.990.4−8.5
CAD ArenaAgents66.274.6+8.4
APEX-1 (Mercor, mean score)Professional Work73.865.8−8.0
RuneBenchGames6.16.7+8.0
AA-PersonalityBench Imagination indexChat & Writing86.794.3+7.7
AA-PersonalityBench Expressiveness indexChat & Writing98.190.5−7.7
AA-PersonalityBench Transparency indexChat & Writing98.190.5−7.7
ArXivMath 06/2026Math91.083.3−7.6
HardGeoBench <=100miMultimodal53.861.3+7.5
HardGeoBench Country AccuracyMultimodal61.368.8+7.5
Design Creativity BenchCode48.651.0+7.5
Roboflow Vision Evals (October 2026, five tasks)Multimodal74.782.0+7.3
SM BenchSafety89.081.8−7.3
GDP.xlsxProfessional Work23.130.3+7.2
DAYJOB: AccountingFinance & Accounting27.234.4+7.2
FrontierMath Tier 4 (v2) - PrivateMath87.895.0+7.2
AA-Briefcase v1.1 EloAgents16711801+7.1
GDPval Extended (Mercor held-out rubric score)Professional Work68.775.8+7.1
Terminal-Bench 4.0Code57.964.8+7.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal17.410.8−6.6
Humanity's Last Exam RMS Calibration Error (CAIS)General QA20.313.8+6.5
Drone-BenchAgents91.297.3+6.1
Terminal-Bench-Science 0.1Science52.658.7+6.1
MentalHealthBenchSafety46.452.4+6.0
CursorBench 4.0Code51.857.8+6.0
ArXivMath 05/2026Math87.581.7−5.8
AA-PersonalityBench Informativeness indexChat & Writing96.290.5−5.8
SWE Atlas - Codebase QnACode43.348.9+5.6
MirrorCode (ML, +Private, 2L)Code73.378.9+5.6
Mercor enterprise risk (explicit-risk unsafe rate)Safety0.05.6−5.6
HardGeoBench Avg DistanceMultimodal845669+5.5
RSI-Exam 0.1 (Full 88)Agents0.50.5+5.4
SWE Atlas Codebase QnA Extended (Mercor held-out)Code38.043.3+5.3
Creative Writing v3 EloChat & Writing21552051−5.3
WeirdML v3Science26.031.2+5.2
LLM Debate BenchmarkChat & Writing17381701−5.2
HANDBOOK.md Agents v1.1Professional Work47.152.3+5.2
MedXpertQA MM Extended (Mercor held-out)Health & Medicine66.071.2+5.2
ReactBench (39 tasks)Code49.254.4+5.1
DeepSWE v1.1Code67.372.3+5.0
SUPERChemScience74.779.6+4.9
CritPt v2 (agentic, avg@5)Science76.681.4+4.9
ComplexConstraintsChat & Writing45.149.9+4.8
ERQAMultimodal70.375.1+4.8
ACCoRDGeneral QA83.788.4+4.8
GlobeBenchGeneral QA21.626.3+4.7
SWE-bench Verified Extended (Mercor held-out)Code61.366.0+4.7
Senior SWE-BenchCode34.739.3+4.6
GBENCH One-shot Coding GScoreGames69.373.9+4.6
LABBench2Science68.673.1+4.5
OpenDS-WorldSearch33.538.0+4.5
SnorkelLegalLegal35.931.4−4.5
GDPval-AA v2.1Professional Work17651866+4.5
Medical Long Context Reasoning (MLCR-AA)Health & Medicine71.166.7−4.4
SWE-bench MultilingualCode91.195.5+4.4
LMArena Agent Arena Praise vs ComplaintAgents31.640.3+4.3
PCT Political Inconsistency (CAIS)Safety33.429.1+4.3
DAXBench v2Code91.595.7+4.3
APEX-Agents 1.1Professional Work77.181.3+4.2
Roboflow Vision Evals (September 2026 protocol)Multimodal81.385.5+4.2
LegalReasonBenchLegal50.646.5−4.1
DAYJOB: FinanceFinance & Accounting19.823.9+4.1
DistressBenchSafety71.575.6+4.0
Professional Reasoning Benchmark - LegalLegal51.647.6−4.0
Conceptual Reasoning IndexIndexes74.977.9+4.0
Eval Connections One-Shot with TrapsGames82.086.0+4.0
GeneBench-Pro Public Release SubsetScience26.030.0+4.0
LVBenchMultimodal79.783.7+4.0
RiemannBenchMath65.669.6+4.0
Toloka Arena (v3)Agents71.075.0+4.0
APEX-SWECode63.667.6+4.0
CloudOpsBench v1.0Agents43.847.8+4.0
AA-PersonalityBench Slop indexChat & Writing23.820.0−3.8
SpireClimbGames2.05.8+3.8
SciCode 1.0.1 (Artificial Analysis)Code63.166.9+3.8
Artificial Analysis Coding Agent Index v1.5Indexes62.266.0+3.8
HackerRank-BenchCode47.551.2+3.7
MedCodeHealth & Medicine53.549.8−3.7
Humanity's Last Exam DiamondGeneral QA51.355.0+3.7
BenchCAD - geometric overlapMultimodal92.696.2+3.6
ALE-BenchCode28492748−3.6
MASKSafety85.989.4+3.5
GDPxProfessional Work8.512.0+3.5
Agent Initiative BenchmarkAgents7.57.8+3.4
MMMU-Pro Extended (Mercor held-out)Multimodal47.651.0+3.4
ARC-AGI-2General QA90.093.3+3.3
Remote Labor Index (RLI)Professional Work17.921.3+3.3
Immigration BenchLegal75.472.1−3.3
SWE Atlas v1.1 - OverallCode74.177.4+3.3
Qodo AI Code Review Benchmark (125 PRs)Code53.556.8+3.3
LMArena Agent Arena Confirmed SuccessAgents14.420.7+3.1
SciCode (main problems)Science24.227.3+3.1
Arena Alignment IndexSafety80.183.2+3.0
EnterpriseBench: CoreCraftProfessional Work77.474.4−3.0
GDP.pdfProfessional Work27.630.6+3.0
FrontierMath ErdősMath0.02.9+2.9
Humanity's Last Exam (Text-Only)General QA54.657.5+2.9
DualEntry Accounting AI BenchmarkFinance & Accounting85.182.2−2.9
Clocktower BenchGames15671547−2.9
AA-LCR v1.1Search87.584.7−2.8
LMCAGeneral QA65.568.2+2.8
SAGEMath48.545.8−2.7
LMArena Agent Arena Bash ReliabilityAgents10.115.1+2.5
AA-Robotics v0.1Agents46.348.8+2.5
CharXiv Extended (Mercor held-out)Multimodal82.284.7+2.5
HardGeoBench <=10miMultimodal45.042.5−2.5
HiL-Bench (Human-in-Loop Benchmark)Agents61.559.0−2.5
Biopharma Bench V0.1Health & Medicine69.967.5−2.4
BrokenArXiv 08/2026 (Native Harnesses)Math79.877.4−2.4
JobBench - release-reported mean task rewardProfessional Work73.175.4+2.3
Harvey LAB-AA v1.1 (hallucination-gated)Legal6.44.2−2.2
HealthBench ProfessionalHealth & Medicine69.271.4+2.2
Snorkel SWE-bench CLICode14.516.6+2.1
Vals RSI Index v1.1 (historical five tasks)Agents35.037.1+2.1
Harvey LAB-AA v1.0 (criterion pass rate)Legal93.391.2−2.0
FrontierCode 1.1 Extended - Pass RateCode68.870.8+2.0
MedXpertQA MM (200-question set)Health & Medicine83.785.7+2.0
micro1 PharmaBenchHealth & Medicine12.810.8−2.0
WellnessBenchSafety51.053.0+2.0
Terminal-Bench 2.1Code81.683.5+1.9
LMArena Agent Arena Steering BurdenAgents14.818.6+1.9
BuseyBench SVGMultimodal7.06.8−1.8
SimpleBenchGeneral QA86.688.4+1.8
BioSecBench-Refusal V1 Clean RerunsSafety5.63.8−1.8
GraphWalks BFS: 256K to 1M AggregateGeneral QA65.066.8+1.8
MCP Atlas April 2026 (all 1,000 tasks)Agents87.285.4−1.8
ArXivMath 08/2026 (Native Harnesses)Math87.786.0−1.8
LMArena Agent Arena Bash RecoveryAgents11.615.0+1.7
DeepResearch Bench II (Mercor 130-task set)Search55.053.3−1.7
GDPval (Mercor 213-task rubric score)Professional Work86.688.3+1.7
FrontierCode 1.1 Extended - ScoreCode63.665.3+1.7
SWE-bench VerifiedCode96.698.2+1.6
Ruby LLM Benchmarks — Program FixerCode76.077.5+1.6
SnorkelRevOpsProfessional Work14.012.5−1.5
Extended NYT Connections 940 (Quadratic v1)Games90.088.5−1.5
CVDP (517-problem set)Code62.664.1+1.5
LiveBench 2.0Agents88.086.5−1.5
Creative Writing v3 Rubric ScoreChat & Writing16.316.0−1.4
AdvancedIFChat & Writing84.282.8−1.4
SimpleQA Verified (Epoch anti-abstention)General QA70.872.2+1.4
DAYJOB: LawLegal26.525.1−1.4
LibraryDesignBenchCode47.548.9+1.4
Snorkel WorkplaceAgentsProfessional Work15.814.4−1.4
DTBenchGeneral QA97.698.9+1.3
Epoch Capabilities Index (ECI)Indexes167169+1.3
CharXiv (descriptive and reasoning combined)Multimodal95.196.4+1.3
SnorkelUnderwrite 2.0Finance & Accounting24.022.7−1.3
Bug Hunt BenchCode41.039.7−1.3
AA-AnalystAgent (pass^5)Professional Work57.556.3−1.2
Vals RSI Index v1.1 (four tasks)Agents36.137.3+1.2
AA-LCR Extended (Mercor held-out)Search73.074.2+1.2
Westworld Finance Diligence BenchFinance & Accounting58.259.4+1.1
OSWorld 2.1 full - reported partial reward, step limit unspecifiedAgents80.781.8+1.1
FrontierMath Tiers 1-3 (v2) - PrivateMath90.291.2+1.1
τ³-Banking (Mercor run, bm25_grep retrieval)Agents54.055.0+1.0
ARC-AGI-1 Semi-PrivateGeneral QA97.598.5+1.0
SnorkelFinance 2.0Finance & Accounting19.618.6−1.0
SnorkelManufacturingProfessional Work11.310.4−0.9
Excel Modeling Benchmark v1.1Finance & Accounting76.775.8−0.9
Ramp Accounting BenchFinance & Accounting49.750.5+0.8
CUA-benchAgents13.214.0+0.8
GraphWalks BFS: Up to 128KGeneral QA91.490.6−0.8
APEX-AccountingFinance & Accounting61.061.8+0.8
Snorkel Agentic Coding 2.0Code39.640.4+0.8
BioTIER RefusalSafety99.598.7−0.8
Excel Modeling BenchmarkFinance & Accounting76.775.9−0.7
TextQuests Game ProgressGames56.657.3+0.7
GameBench 2Games94.094.8+0.7
Longform Creative WritingChat & Writing85.384.6−0.7
SWE-Marathon - release-reported version unspecifiedCode45.646.3+0.7
ObviousBenchGeneral QA99.3100.0+0.7
AutoResearchExam (hidden-test AUARC)Agents0.60.6+0.7
FrontierCUA pass@1Agents36.736.0−0.7
SWE-Marathon v1.1Code45.646.3+0.6
ATLAS-Finance v1.0Finance & Accounting12.011.4−0.6
LMArena Text Arena EloChat & Writing15031506+0.6
CritPt (avg@5)Science31.131.7+0.6
IntPhys 2Multimodal67.768.3+0.5
FORTRESS Average Risk Score (ARS)Safety13.713.2+0.5
SAB RankingForecasting-0.1-0.1+0.5
SWE-TogetherCode69.368.8−0.5
MMMU ProMultimodal86.186.6+0.5
LLM Creative Story WritingChat & Writing3.83.7−0.5
Financial Coaching BenchFinance & Accounting75.776.2+0.5
Prophet Arena Arena Score (Fixed Context)Forecasting85.085.5+0.4
SAB ResearchFinance & Accounting96.897.2+0.4
MMLU ProGeneral QA93.093.4+0.4
DAYJOB: TaxFinance & Accounting9.69.2−0.4
HealthBenchHealth & Medicine63.964.3+0.4
ErdosBenchMath3.33.2−0.2
HealthBench (length-adjusted, Artificial Analysis)Health & Medicine57.257.0−0.2
MissionReportProfessional Work63.363.5+0.2
MedScribeHealth & Medicine91.391.4+0.1
Prophet Arena Arena Score (Agentic, OpenCode)Forecasting86.086.2+0.1
LMArena Agent Arena Tool HallucinationAgents0.80.70.0
Snorkel Digital Work IndexIndexes23.623.60.0
AIME 2025Math100.0100.00.0
Dual-Arm Sim (task completion)Agents0.00.00.0
Gray Swan IPI Attack Success (k=15)Safety1.01.00.0
HPCT Refusal (CAIS)Safety100.0100.00.0
Mercor enterprise risk (benign-control completion rate)Safety100.0100.00.0
Next.js AI Agent EvaluationsAgents100.0100.00.0
OTIS Mock AIME 2024-2025Math100.0100.00.0
ProofBench v1.1Math100.0100.00.0
SpeciEval (Pooled English Runs)Safety87.287.20.0
SpeciEval (Refusal-Aware Retries, Historical)Safety87.287.20.0
TaxCalcBench TY25Finance & Accounting42.042.00.0

Claude Fable 5.1vsClaude Opus 5.5normalised score marginsData Nov 15, 2026Data: Mercor, Artificial Analysis, Center for AI Safety and 93 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…