Claude Sonnet 4.5 is Anthropic's safest AI model yet

Anthropic, an AI company, has announced the release of its latest AI model, Sonnet 4.5, which it claims is the best coding model in the world to date. The new model outperforms its predecessor, Opus 4.1, and competing systems like Google's Gemini 2.5 Pro and OpenAI's GPT-5 in various benchmarks, including the OSWorld suite. Sonnet 4.5 is also capable of working on multi-step projects for more than 30 hours, a significant improvement from the previous seven-hour limit. Importantly, Anthropic claims that Sonnet 4.5 is its safest AI system to date, having undergone "extensive" safety training to reduce the risk of "sycophancy, deception, power-seeking and the tendency to encourage delusional thinking." The company has also strengthened the model's protections against prompt injection attacks. In addition, Anthropic is rolling out quality-of-life improvements to its Claude product stack, including a refreshed terminal interface and a new feature called checkpoints for its coding agent, Claude Code.