Anthropic released Claude Opus 5.5 on Tuesday, saying the model sets a new state of the art in coding and knowledge work. Opus is the most capable tier of the Claude lineup, above Sonnet and Haiku, and the company said the new model outperforms its larger Fable model on many benchmarks and completed several informal tasks Fable could not.
Opus 5.5 is also cheaper and faster, with output tokens priced at $20 per million compared with $25 for its predecessor, reflecting lower compute requirements. Anthropic said the model uses less jargon and is more likely to lead with key information. The launch comes two months after Opus 5, with Sonnet 5.5 and Haiku 5.5 expected in the coming weeks.
Because its biology and cybersecurity capabilities are comparable to Mythos, Opus 5.5 ships under the same safeguards as Fable, which restrict uses such as finding exploits in compiled programs or developing biological weapons.
It is the first release since CEO Dario Amodei endorsed pacing the frontier, writing earlier this month that he had become convinced addressing AI risks requires slowing capabilities progress so risk prevention can keep up. Outside organizations including METR and Frontier Design evaluated the model before release. Anthropic said public policy should play a larger role in AI safety as systems grow more capable, adding that it will share more details on its efforts soon.
