Claude Opus 4.5

anthropic · Ranked across 4 benchmarks · best rank #1

Benchmark scores

BenchmarkCategoryRankScoreCaptured
SWE-bench Verified agents #1 79.2% 2025-12-15
Terminal-Bench 2.0 agents #7 58.4% 2026-01-16
Chatbot Arena chat #12 1448 2026-04-30
OpenRouter · Weekly Usage usage #24 #64 2026-05-02