Arena.ai says Claude Opus 5.5 is easier to read than Opus 5 but writes longer answers

Arena.ai analyzed how Claude Opus 5.5 writes compared with Opus 5 across high-reasoning Text Arena outputs. It says 10 of 12 writing measures moved in a better direction, and that Opus 5.5 should be easier to read.

Long content words fell from 41.7% to 38.6%, the lowest share of any Claude model in the analysis. Average sentence length dropped 17%, from 12.14 to 10.03 words. Answers became 6% longer, rising from 453 to 481 words on average. Arena.ai says that makes Opus 5.5 the longest-answering model in the Opus family.

Arena.ai says the model also sounds less recognizably AI on two familiar tells, with 95% fewer em dashes and 73% fewer semicolons. It says a new giveaway may be emerging: hedges and caveats such as “perhaps” and “arguably” rose 97%, from 0.39 to 0.77 per 1,000 words, the highest rate of any Claude model in the analysis.