Pinned
We evaluated @claudeai Opus 5.5 on Box's Complex Work Eval. The standout finding is efficiency.
→ 30% faster than Opus 5
→ A third of the tokens
→ Answers 42% shorter with the findings intact
What disappeared was restatement and preamble. Opus 5.5 simply led with the
