Anthropic released Claude Opus 5.5 as the inaugural model in its new 5.5 family. The model matches the performance of the top-tier Fable 5.1 across most tasks. Operating costs for Opus 5.5 are approximately 40% lower than the previous Opus 5. The system operates over 30% faster than its predecessor. Efficiency improvements also allow the model to require fewer tokens to produce high-quality work.
This launch represents Anthropic’s first release since co-founder Dario Amodei called to pace the frontier of AI development. External evaluators audited the model, which achieved the company’s highest internal behavioral scores to date. Enhanced safeguards reduce risks of prompt injections and attempts to escape testing environments. The system may redirect sensitive cybersecurity queries to less powerful models to mitigate potential risks.