Opus 5.5 Sets New Terminal-Bench Bar

The score gives teams a sharper reference point for judging terminal-based coding agents, not another vague claim of model intelligence.

AI-written, human-edited, never fabricated. How this is made