swyx says Claude Opus 5.5 is AINews' default model, writing tighter summaries than Opus 5

Swyx, who runs AINews, says Claude Opus 5.5 has become the default model for the newsletter because its reporting is more concise and tasteful, with less filler than Opus 5.

He says a side-by-side test against 6 Sol showed a large difference in favor of Opus 5.5, though the report gives no output samples or scoring method for that comparison.

Swyx also reports a complaint: Opus performs poorly when prompts contain certain banned words. The report does not name the words or say how often the problem occurs.