Claude Fable 5.1 (xHigh): benchmark results

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

Unified ELO 1792 ± 1, rank #9 of 1761 rated models, from 9 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
SEAL - Humanity's Last Exam46.5Score100
Epoch AI - Critpt31.14Score99.4
CursorBench 3.172.8Score (%)98.6
SEAL - Humanity's Last Exam (Text Only)46.8Score98.3
ARC-AGI-290Accuracy (%)96.8
ARC-AGI-196.5Accuracy (%)93.6
LLM2014 Logic 2026-0963.47Median Score90.9
SEAL - Professional Reasoning Benchmark - Legal51.6Score87.1
SEAL - Professional Reasoning Benchmark - Finance50.8Score80.6

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