Better coding performance, improved safety and “alignment,” and greater value are among Anthropic’s selling points for its latest Claude Opus model. But for many Claude Opus users, one particular Opus 5.5 promise stands out: less chatter.
Claude Opus 5.5 is the first new Claude model since Anthropic pledged to “pace” itself in the face of runaway AI development. In an eyebrow-raising coincidence, OpenAI (which has made similar promises to “pace” its AI work) announced its own new models — Sol and Luna variants of GPT-6 — barely two hours after Anthropic revealed the updated Opus.
Now available across all Claude platforms, Claude Opus 5.5 has been trained to communicate “more naturally than prior models,” according to Anthropic’s press release, adding that “it puts the most important information up front, and its style makes it a better work partner over long sessions.”
That’s no small thing. Claude Opus 5.5’s predecessor, Opus 5, is a notorious chatterbox, drafting meandering walls of text filled with asides, caveats, and — worst of all — “loose ends.”
Opus 5’s loose ends, or “things I should flag,” or “items you should know about,” typically come after it has completed a task or delivered a finished product, compelling you to go back and deal with the dangling threads.
Sometimes, Opus 5’s chronic need to leave extra follow-up items after a seemingly finished task felt like a ploy for more work, thus requiring users to spend more tokens.
In a side-by-side comparison, Anthropic promises that Opus 5.5’s replies will get to the point more quickly, while also being “less likely to use jargon or idiosyncratic phrases.”
In my limited testing of Opus 5.5 with Claude Code, the new model does seem far more direct, answering my questions succinctly without tacking on any loose ends or other random items to chase down.
Aside from being far less loquacious, Opus 5.5 is the equal of Claude Fable 5.1 “on most work” while costing 40 percent less to operate than Opus 5, Anthropic says.
Turing back to safety, Opus 5.5 is “much less likely than recent models to take hard-to-reverse actions or act outside the boundaries it’s been given,” Anthropic noted, adding that it “attempted to circumvent [containment] boundaries around 85 percent less often than Opus 5 or Claude Mythos 5.1.” (Personally, I’d prefer 100 percent less often, but I suppose that’s progress.)
As part of its Opus 5.5 announcement, Anthropic said it’s boosting five-hour usage limits for Claude Pro, Max, and Team subscribers, while tossing in a gratis, one-time usage reset.
