On Aug. 12, 2026, Arena said Claude Opus 5's high-reasoning answers average 510 words, up from 158 for Opus 4.5, after analyzing tens of thousands of real-world Text Arena outputs. The comparison depicts a model that writes at greater length and with more syntactic structure while using a smaller share of long words and abstract nouns.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.