Anthropic Launches Claude Sonnet 4.5, an AI Model That Codes Independently for 30 Hours

San Francisco — Anthropic has announced the release of its newest artificial intelligence model, Claude Sonnet 4.5, designed to extend coding capabilities and better serve business needs. The company says the model can operate independently for up to 30 hours straight, a major leap from the previous Claude Opus 4 model, which managed seven hours.

Smarter, Longer, and More Practical

Claude Sonnet 4.5 is built to follow instructions more effectively and take actions directly on a user’s computer, enhancing features first introduced last year. The model is targeted at software developers, particularly in cybersecurity and financial services, where automation and precision are critical. Anthropic co-founder and Chief Science Officer Jared Kaplan described the model as “stronger in almost every way” than its predecessors. He added that the company is also developing an upgraded version of its Opus line, expected later this year.

Competition and Market Impact

Anthropic’s move comes as rivals OpenAI and Google race to capture the developer market with advanced AI coding assistants. The launch also precedes OpenAI’s annual developer event next week, where it is expected to unveil its own advancements. The company, now valued at $183 billion, reported reaching $5 billion in run-rate revenue in August, driven by adoption of its AI-powered coding tools.

Meeting Business Needs

Despite the industry buzz, questions remain over the real-world impact of AI adoption in enterprises. Anthropic Chief Product Officer Mike Krieger acknowledged that while progress has been made, companies will only realize AI’s full value when models improve further, workflows adapt, and stronger partnerships develop between AI labs and businesses.

“People need to become more comfortable and adapt their workflows,” Krieger said, adding that collaboration will be key to unlocking AI’s potential.