Claude Opus 5.5 Tops Coding Agent Index but Costs 21% More Per Task

Artificial Analysis has launched its Coding Agent benchmark leaderboard, where Claude Opus 5.5 took the top spot with a score of 66 — the highest they've ever measured — even as token consumption and real cost per task both went up.

Confirmed

Why it matters

2026-09-23 ~ 2026-09-24 · 5 related posts

Full story(12 episodes)→

Primary sources