Anthropic released Claude Opus 5.5 on Tuesday, September 22, and it is in the Chat model menu now.
What Anthropic says changed
Opus 5.5 is the first model in the Claude 5.5 family. Anthropic says it performs at the level of Claude Fable 5.1 on most tasks, while costing about 40 percent less to run than Claude Opus 5 on typical workloads. The list price is $4 per million input tokens and $20 per million output tokens, 20 percent below Opus 5. Anthropic also reports faster output, more than 30 percent quicker than its predecessor.
On coding in a terminal, Anthropic reports a Terminal-Bench 4.0 score of 66.4 percent for Opus 5.5, against 55.8 for Fable 5.1 and 52.3 for Opus 5. On computer tasks it reports 81.8 percent on OSWorld 2.1, up from 74.0 for Opus 5. These are the company's own figures from its launch post, and they have not been independently repeated. A chart of the Terminal-Bench scores is in Charts in Chat.
Anthropic also describes stronger safeguards. For a small set of requests it considers sensitive, in cybersecurity and biology, the model can fall back to an earlier one. If a reply seems more cautious than expected on a topic like that, the fallback is a likely reason.
Using it in Chat
Choose Claude Opus 5.5 from the model menu before sending a message. Because the model is picked per message, a conversation can start on a faster, cheaper model and move to Opus 5.5 for the one question that needs it. The model that answers next reads everything said so far, and nothing has to be pasted across. Chat on the web and iPhone explains how switching works.
Two things are worth keeping in mind. A lower price per token does not guarantee a lower price per task, since a model that thinks longer uses more tokens. And Opus 5.5 draws on the same monthly allowance as every other model on a plan.
