The AI Race Goes Into Higher Gear. Anthropic’s Claude Sonnet 4.5 Is a Game-Changer.

Claude Sonnet 4.5 can take directions and complete multistep tasks. (Anthropic)

The AI model arms race just went to another level.

On Monday, Anthropic introduced its newest artificial-intelligence model, called Claude Sonnet 4.5.

”Claude Sonnet 4.5 is the best coding model in the world. It’s the strongest model for building complex agents,” Anthropic said in a blog post. “It’s the best model at using computers. And it shows substantial gains in reasoning and math.”

The AI startup said Sonnet 4.5 will be available to Claude users on Monday.

Many technology executives expect the advent of AI agents, which are programs that have the ability to take simple directions and complete multistep tasks, to drive the next wave of technology industry growth.

Anthropic claims Sonnet 4.5 can work on complicated multistep coding tasks for more than 30 hours at a time. The model achieved a new record high in the coding benchmark SWE-bench, a widely used industry benchmark to evaluate the performance of AI models on real-world software engineering tasks.

The company also said customers working in several fields—including finance, law, medicine, and science—have found a preview version of Sonnet 4.5 has much-improved “knowledge and reasoning” capabilities versus older models.

“Claude Sonnet 4.5 is state of the art on the most complex litigation tasks,” Thomson Reuters executive Pablo Arredondo said.

Earlier this month, Anthropic announced that it raised $13 billion in new capital that valued the startup at $183 billion. At the time, the company said its revenue had reached an annualized run rate surpassing $5 billion with more than 300,000 business customers, up from $1 billion annualized revenue run rate at the beginning of the year.

Write to Tae Kim at tae.kim@barrons.com