One tester handed Claude Opus 5.5 a 680,000-line code migration and had it done in under a day. An engineering team would have needed weeks. That result sits at the center of Anthropic’s launch of the model, the first in its new Claude 5.5 family.
Anthropic says Opus 5.5 performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5. Outside evaluators, including Frontier Design and METR, tested it before release.
What it can do
Think of it like a contractor who finds the slow parts of your house and fixes them without knocking down a wall. Anthropic asked the model to cut load times on every page of a web app. It succeeded 39 times out of 40. Opus 5 made smaller gains and also changed how the app behaved.
In another test, a tester had several Claude models build a game from a single prompt. Opus 5.5 scored higher than any other on graphics and polish. On Anthropic’s own benchmarks it leads in agentic coding, computer use and knowledge work, though the company warns that at this level, benchmark margins say less about real-world differences than they used to. In its own use, the gap between Opus 5.5 and Fable 5.1 is smaller than the scores suggest.
Safety
On Anthropic’s automated behavioral audit, the most comprehensive alignment test it runs, Opus 5.5 posted the best scores of any model so far. It is much less likely than recent models to take actions that are hard to undo or step outside the limits it was given, and it resists prompt injection better than Opus 5. Anthropic widened the testing to cover longer tasks, impossible tasks and scenarios modeled on real incidents. It says the model still has limits. Full details are in the Opus 5.5 System Card.
Opus 5.5 matches Claude Mythos 5.1 in biology and cybersecurity, so it ships with safeguards like those on Fable 5.1. Vetted organizations can apply now to the Life Sciences Verification Program for biology research. Access to the Cyber Verification Program will widen in the coming weeks for verified security practitioners.
Price and speed
The practical upshot is cost. Input and output tokens run $4 and $20 per million, 20% below Opus 5. Cache reads, which make up most of the bill in agentic and coding work, are $0.20 per million, 60% below Opus 5. Output arrives more than 30% faster. Anthropic also raised five-hour usage limits on Pro, Max, Team and seat-based Enterprise plans, and subscribers now get a rate limit reset they can save and use when they choose.
Writing and what is next
Early testers said Opus 5.5 writes more clearly and is easier to follow, which answers a common complaint about Opus 5. “It writes the way I do,” one of them said. Anthropic adds that clearer output is easier to check, which helps safety as well as convenience.
Claude Sonnet 5.5 and Claude Haiku 5.5 are due in the coming weeks with many of the same improvements.
