Claude Opus 4.5

About

The page announces Claude Opus 4.5, Anthropic's newest and most intelligent AI model, available today. It is highlighted as the best model for coding, agents, and computer use, and significantly improved for everyday tasks like deep research and working with slides/spreadsheets. The model is state-of-the-art on real-world software engineering tests (SWE-bench Verified) and is available via apps, API, and major cloud platforms with competitive pricing. The announcement includes testimonials from early access customers praising its reasoning, efficiency, code quality, and ability to handle complex, long-horizon tasks. It also details evaluation results, showing it outperforms human candidates on internal exams and leads on various programming and agentic benchmarks. A key feature is its creative problem-solving, demonstrated by an example of navigating airline policy. Furthermore, the page emphasizes significant advancements in safety, making Opus 4.5 the most robustly aligned model and highly resistant to prompt injection attacks. Updates to the Claude Developer Platform, Claude Code, and consumer apps are also mentioned.