Claude Opus 5.5 lands on Amazon Bedrock, Claude Platform on AWS

Claude Opus 5.5 arrives as the first member of the Claude 5.5 model family, and AWS is making it available through its own platform rather than only through Anthropic's direct channels. The body frames the model around three jobs — agentic coding, knowledge work, and long-running tasks — and pairs that with practical guidance for developers on Amazon Bedrock.

The efficiency story is central. Anthropic says Opus 5.5 does more with fewer tokens than Claude Opus 5, and AWS writes that lower per-token prices and much cheaper cache reads stack on top of those gains, producing an average lower cost per task. Alongside cost, the model is trained to communicate more clearly, surfacing what it did, what it found, and what it needs as it works, while adaptive thinking is always on and effort replaces manual thinking budgets as the control.

Safety is a notable trade-off. Opus 5.5 is the first Opus model with safety classifiers similar to Claude Fable 5.1 in biology, cyber security, and AI development, and the body notes requests will be refused more frequently than with previous Opus versions. Whether that trade-off pays off hinges on whether the clearer communication and lower cost per task outweigh the added refusals for the developers and knowledge workers running long sessions on Amazon Bedrock.