Anthropic launches Claude Opus 5 AI to find software bugs

ANTHROPIC has launched Claude Opus 5, a more affordable alternative to its high-end Fable 5 AI model. Opus 5 is particularly effective at identifying software vulnerabilities, nearly matching the capabilities of Mythos 5 but falling short in exploit development. The model intentionally avoids direct training on offensive cyber tasks, leading to a lower exploit development score. Safety features in Opus 5 are less restrictive than those in Fable 5, allowing more freedom in searching source code for vulnerabilities. It is available for $5 per million input tokens and $25 per million output tokens, matching the pricing of Opus 4.8.