Claude Opus 5.5 shows sharp drop in em dashes and shorter sentences, Arena finds
Benchmarking site Arena analyzed high-reasoning Text Arena responses from August and September 2026 and found Anthropic's Claude Opus 5.5 uses far fewer em dashes and semicolons than Opus 5, along with shorter, simpler sentences. Em dash usage fell from 15.2 to 0.8 per 1,000 words, and average sentence length dropped from 12.14 to 10.03 words, though overall answers grew longer, from 453 to 481 words on average.
GoKawiil's interpretation of the reporting above, not reported fact.
Em dashes and semicolons have often been cited as telltale signs of AI-generated text, so a sharp reduction could make Claude's output harder to flag as machine-written, which may affect how people detect AI slop online. The added verbosity suggests Anthropic may be trading brevity for other qualities, though the underlying reasons for the stylistic shift are not stated in Arena's findings.
- Opus 5.5 cuts em dash usage by about 95% compared with Opus 5, per Arena's analysis.
- Sentences are shorter and simpler, but overall responses are longer than any previous Opus model.
- 10 of 12 writing measures tracked by Arena moved in a direction it considers improved.
Source: bleepingcomputer.com, 2026-09-26
Published there as: “Claude Opus 5.5 uses 95% fewer em dashes, but its answers are getting longer”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.