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

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

Unified ELO 1747 ± 1, rank #34 of 1761 rated models, from 19 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Humanity's Last Exam55.93Accuracy (%)99.7
AA Omniscience - Business55.8Accuracy (%)99.4
AA Omniscience - Health55.5Accuracy (%)99.4
AA Omniscience - Humanities & Social Sciences63.1Accuracy (%)99.4
AA Omniscience - Law66.1Accuracy (%)99.4
AA Omniscience - Science, Engineering & Mathematics60Accuracy (%)99.4
AA-Omniscience Accuracy64.93Accuracy (%)99.4
AA CritPt30.29Accuracy (%)99
AA Omniscience40.8Score98.6
AA Omniscience - Software Engineering (SWE)89.1Accuracy (%)98.5
AA Long Context Reasoning83.67Accuracy (%)98.3
Artificial Analysis Intelligence Index51.67Intelligence Index98.2

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