Claude Opus 5.5 (Max): benchmark results

Provider: Anthropic. Access: API.

Unified ELO 1787 ± 1, rank #9 of 3363 rated models, from 37 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
DataBench70.5Score (%)100
LiveBench Olympiad93.25Score100
LiveBench TypeScript63.33Score100
Vals AI MedScribe91.43Accuracy (%)100
Vals AI ProgramBench86.99Raw Pass Rate (%)100
Vals AI RSI Index0.37Index Score (0-1)100
LiveBench JavaScript81.82Score99.2
LiveBench Code Generation91.55Score98.4
LiveBench Integrals With Game100Score98.4
LiveBench Code Completion86.96Score97.6
Bug Hunt Bench - LMS23.5Planted Bugs Fixed (out of 60)97.5
ARC-AGI-291.67Accuracy (%)96.6

Interactive version: theaggregate.ai/model?slug=claude-opus-5-5-max · How It Works · Data refreshed daily, snapshot 2026-09-23.