Opus 5.5 writes shorter sentences and 95% fewer em dashes, but hedges rise 97%
arena · x · 2026-09-26
Text Arena compared how Opus 5.5 writes versus Opus 5 across high-reasoning Text Arena outputs, finding 10 of 12 writing measures improved:
- Long content words fell from 41.7% to 38.6%, the lowest of any Claude model analyzed
- Sentences are 17% shorter on average (12.14 → 10.03 words)
- Classic AI tells dropped sharply: 95% fewer em dashes and 73% fewer semicolons
- The tradeoff is length: average answers grew 6% (453 → 481 words), the longest in the Opus family
- A new tell may be emerging: hedges like "perhaps" and "arguably" rose 97% (0.39 → 0.77 per 1,000 words), the highest rate of any Claude model analyzed
Related event: Text Arena: Opus 5.5 Writes More Like Humans, Em Dash Use Down 95%(5 posts)→
More from Models
- Micius-Flash: 27B Multimodal Model for Hardware and Semiconductor Agents — zhiman-ai · 2026-11-11
- Claude IP risk-check site goes viral as users audit their IPs ahead of new rules — vista8 · 2026-11-11
- Baidu says RSI auto-optimized ERNIE inference, 6.24x decode speedup on Kunxin P800 — xiaohu · 2026-11-11
- Baidu's ECHO: multi-token prediction reusing the last layer as a drafter — xiaohu · 2026-11-11
- ERNIE 5.5 Flash Preview: 280B-param multi-router MoE activating only 14B per token — xiaohu · 2026-11-11
- ERNIE 5.5 recreates the Pelican test: 117-second animation drawn frame-by-frame in Python — xiaohu · 2026-11-11