Anthropic launches cheaper Claude Opus 5.5 with tighter safety controls

Anthropic has released Claude Opus 5.5, a new AI model aimed at software development and complex office work. The company says it matches the performance of its more powerful Fable 5.1 model on most tasks while costing 40% less to run than Opus 5, a meaningful change for companies that pay by the amount of text an AI system processes and generates. The tradeoff is that Anthropic is treating the model as powerful enough in cybersecurity and biology to warrant the stricter controls used for its frontier models: access for certain sensitive work is restricted to vetted researchers and security professionals. Anthropic says independent groups tested it before launch and that it is harder to trick into unsafe actions, though those evaluations are the company’s own safety case rather than proof that risk has disappeared.