Atlas

Compare Models

Claude Fable 5 vs Claude Fable 5.2

Claude Fable 5.2 leads by 6.8 capability points (≈95% interval +4.0 to +9.6): a clear lead over Claude Fable 5 · P(higher) ≈ 100%

Wins 42 of 45 shared benchmarks (2 lost, 1 tied)

Capability vs. relative cost
Claude Fable 5Claude Fable 5.2Closed 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 5vsClaude Fable 5.2capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins2 – 42 · 45 shared, 1 tied

Comparing matching setups only (11 benchmarks excluded) · suites counted once
← Claude Fable 5Claude Fable 5.2 →

Claude Fable 5vsClaude Fable 5.2wins by categoryData Nov 15, 2026Data: Epoch AI, Arena, Anthropic and 15 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Claude Fable 5.2 minus Claude Fable 5.

Anthropic ECI (AECI), November 2026 fitIndexes169177+74.0
Furniture Assembly BenchmarkMultimodal35.890.0+54.2
LMArena Agent Arena Net Improvement (Steering Burden)Agents7.330.4+43.5
ChartographyMultimodal34.873.4+38.6
Mystery Game PuzzlesGames52.090.0+38.0
Humanity's Last Exam DiamondGeneral QA45.680.2+34.6
AutomationBench - release-reported metric and version unspecifiedAgents17.151.4+34.3
Clocktower BenchGames15341759+32.3
Chess Puzzles (task version 2.0)Games47.079.0+32.0
LMArena Agent Arena Praise vs ComplaintAgents14.373.4+29.6
Bug Hunt BenchCode27.655.2+27.6
LMArena Code Arena WebDevCode16251836+27.5
AutomationBench 1.0.6 (Zapier)Agents17.144.3+27.2
GameBench 2Games75.299.1+23.9
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine20.643.9+23.3
HieroglyphBenchMultimodal22.943.6+20.7
SWE-bench MultimodalCode54.974.1+19.2
InferenceBenchCode8.710.3+18.3
Humanity's Last ExamGeneral QA57.873.2+15.4
AA-Briefcase v1.1 EloAgents15361785+13.7
Vals RSI Index v1.1 (four tasks)Agents24.337.9+13.6
APEX-Agents 1.1Professional Work74.587.6+13.1
SWE-Bench Pro (Public Dataset)Code80.393.0+12.7
LMArena Agent Arena Steering BurdenAgents7.933.0+12.5
GPQA DiamondGeneral QA83.393.9+10.5
Conceptual Reasoning IndexIndexes72.779.9+9.9
FrontierMath Tier 4 (v2) - PrivateMath90.2100.0+9.8
LMCAGeneral QA61.170.6+9.5
APEX-Accounting 1.1Finance & Accounting61.070.4+9.4
LMArena Agent Arena Confirmed SuccessAgents5.122.0+8.4
FrontierMath Tiers 1-3 (v2) - PrivateMath87.095.1+8.1
HealthBench Professional (Length-Adjusted)Health & Medicine60.968.1+7.2
ProofBench v1.1Math95.0100.0+5.0
ARC-AGI-2General QA89.293.8+4.6
ACCoRDGeneral QA85.089.4+4.4
OfficeQAProfessional Work79.083.4+4.4
MineBench — Bradley-Terry RatingGames18691932+3.8
Vibe Code Bench v1.1Code90.493.8+3.4
Tax Agent BenchFinance & Accounting69.867.0−2.8
LMArena Agent Arena Bash ReliabilityAgents10.414.4+2.0
SWE-bench MultilingualCode94.896.6+1.8
DTBenchGeneral QA98.499.2+0.8
LMArena Agent Arena Tool HallucinationAgents0.41.0+0.3
OTIS Mock AIME 2024-2025Math100.099.7−0.3
ARC-AGI-1 Semi-PrivateGeneral QA98.598.50.0

Claude Fable 5vsClaude Fable 5.2normalised score marginsData Nov 15, 2026Data: Epoch AI, Arena, Anthropic and 15 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…