Claude Opus 5 (Adaptive Reasoning, Xhigh Effort): benchmark results

Provider: Anthropic. Released 2026-07-24. Access: API.

Unified ELO 1736 ± 1, rank #46 of 1761 rated models, from 19 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Artificial Analysis Intelligence Index53.37Intelligence Index99.2
AA Humanity's Last Exam54.4Accuracy (%)98.8
AA Omniscience - Science, Engineering & Mathematics56.51Accuracy (%)98.8
AA Omniscience - Business51.5Accuracy (%)98.4
AA GDPval1713.13ELO98.3
AA Omniscience - Health52.42Accuracy (%)98.2
AA GPQA Diamond93.74Accuracy (%)98
AA Omniscience35.38Score97.9
AA-Omniscience Accuracy59.5Accuracy (%)97.7
AA Omniscience - Humanities & Social Sciences57.1Accuracy (%)97.3
AA CritPt27.71Accuracy (%)96.6
AA Omniscience - Software Engineering (SWE)83.7Accuracy (%)96

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