When Palo Alto Networks earlier this year began testing Anthropic’s Claude Mythos to comb through its own source code, it didn’t take long to see the future of cybersecurity. It also didn’t take long to see what that future would cost.

The model found more than two dozen critical vulnerabilities in around three weeks, roughly five times what the company would typically find using existing tools, said Sam Rubin, senior vice president of Palo Alto Networks’ threat intelligence arm. The company also “very quickly” burned through more than $1 million worth of tokens using Mythos, he said.