Which is better, Claude Fable 5 or Claude Fable 5.1?

Refresh Models —Deals —Regulatory —Sources —

Claude Fable 5 vs Claude Fable 5.1

42 SHARED BENCHMARKS

Across 42 shared benchmarks, Claude Fable 5 scores higher on 10 and Claude Fable 5.1 on 32. The widest gap is Terminal-Bench 4.0, where Claude Fable 5.1 scores 56 against 44.5.

ANTHROPICVSANTHROPIC42 SHARED10–32 HEAD-TO-HEAD

BenchmarkClaude Fable 5MARGINClaude Fable 5.1

AA Agentic Index56.661.3

AA Intelligence62.165.7

AA-LCR76.780

AA-Omniscience43.343.5

arc_agi_198.597.5

ARC-AGI-288.390

Artificial Analysis Coding Index76.581.6

critpt28.631.1

DeepSWE 1.17067.4

GDP (Surge AI)87.185.1

gdpval61.167.7

GDPval-AA v217601853

GDPval-AA v2 Elo17471835

GMMLU93.694

GPQA Diamond92.693.7

HealthBench (raw)61.266.7

HealthBench Professional (raw)68.974.2

healthbench_professional6662.1

HLE64.565

HLE (with tools)64.565

livebench_agentic_coding62.266.1

livebench_coding8686.4

livebench_data_analysis80.580.3

livebench_instruction_following75.873

livebench_language90.789.5

livebench_math9697

livebench_reasoning89.791.7

MILU92.993

officeqa_pro69.969

OmniScience Accuracy65.367.2

OmniScience Non-Hallucination36.434.4

OSWorld 2.066.177.9

Program Bench76.887.6

scicode60.262

simpleqa_verified70.770.8

swe_bench_multimodal54.154.7

SWE-bench Multilingual86.689.1

SWE-bench Pro8081.2

TauBench V3 - Banking38.147.2

Terminal-Bench 2.18891.4

Terminal-Bench 4.044.556

Toolathlon Verified77.977.8

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.