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

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

Unified ELO 1801 ± 1, rank #5 of 1761 rated models, from 20 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA GDPval1769.1ELO100
AA Humanity's Last Exam59.13Accuracy (%)100
AA Omniscience - Health57.4Accuracy (%)100
AA Omniscience - Humanities & Social Sciences66.1Accuracy (%)100
AA Omniscience - Science, Engineering & Mathematics62Accuracy (%)100
AA Omniscience - Software Engineering (SWE)92.4Accuracy (%)100
AA SciCode63.08Accuracy (%)100
AA-Briefcase61.52Rubric Pass Rate (%)100
AA-Omniscience Accuracy67.23Accuracy (%)100
Artificial Analysis Intelligence Index56.76Intelligence Index100
Epoch AI - Scicode62.04Score100
AA Long Context Reasoning85.33Accuracy (%)99.8

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