Anthropic launches Claude Opus 5.5, cutting costs by 40%

METR: METR is an independent AI safety research group specializing in model evaluation. It conducted pre-launch testing of Claude Opus 5.5 alongside Frontier Design. The evaluation focused on safety measures and risk mitigation. Anthropic: Anthropic is an AI company developing advanced language models with a focus on safety and responsible deployment. The firm launched Claude Opus 5.5 on September 22 with external safety evaluations and cost-efficient performance improvements. CEO Dario Amodei has publicly advocated for slower capability releases to address broader AI risks. Dario Amodei: Dario Amodei serves as CEO of Anthropic and leads its strategic direction on model development and safety. Earlier in September he called on the global AI community to moderate the pace of new releases amid rising safety concerns. His comments align with the timing of the Claude Opus 5.5 launch. Google Cloud: Google Cloud provides enterprise cloud services and supports AI model deployment. Anthropic made Claude Opus 5.5 accessible on Google Cloud as part of the rollout. Claude Opus 5.5: Claude Opus 5.5 is Anthropic's newly released AI model positioned as a high-performance option with enhanced safety features. It underwent testing by independent research groups and became available on major cloud platforms immediately upon launch. The model is part of a planned series that will soon include Sonnet 5.5 and Haiku 5.5 variants. Frontier Design: Frontier Design is an independent AI safety research organization. It performed external testing on Claude Opus 5.5 ahead of the public release. The group evaluated the model's safeguards and containment properties. Microsoft Azure: Microsoft Azure delivers cloud computing resources and integrates AI models from various providers. Claude Opus 5.5 launched with availability on Azure alongside other major platforms. Amazon Web Services: Amazon Web Services offers cloud infrastructure and hosts third-party AI models for developers. Claude Opus 5.5 is now available through its platform following the September launch. Risk Debate: Public discussion around AI risks has intensified alongside recent model releases from leading labs. AI Safety Evaluation: Independent research groups conducted pre-release testing of the model to validate safeguards. Platform Integration: The new model is accessible across multiple major cloud providers to support broader developer use.