Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback): benchmark results

Provider: Anthropic. Released 2026-09-01. Access: API.

Unified ELO 1778 ± 1, rank #14 of 1761 rated models, from 18 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Omniscience - Law68.6Accuracy (%)100
AA Humanity's Last Exam58.71Accuracy (%)99.8
AA Omniscience - Science, Engineering & Mathematics61.7Accuracy (%)99.8
AA-Omniscience Accuracy66.22Accuracy (%)99.8
AA Omniscience - Health55.9Accuracy (%)99.7
AA GDPval1751.06ELO99.6
AA Omniscience - Business56.3Accuracy (%)99.6
AA Omniscience - Humanities & Social Sciences64.1Accuracy (%)99.6
AA CritPt31.14Accuracy (%)99.4
AA Omniscience - Software Engineering (SWE)90.7Accuracy (%)99.4
Artificial Analysis Intelligence Index53.8Intelligence Index99.4
AA Omniscience42.38Score99

Interactive version: theaggregate.ai/model?slug=claude-fable-5-1-adaptive-reasoning-xhigh-effort-default-fallback · How It Works · Data refreshed daily, snapshot 2026-09-05.