Anthropic has introduced Opus 5, the latest version of its flagship AI model, positioning it as a more flexible and cost-efficient option than Fable 5. Despite being smaller, the new model delivers stronger results in several benchmark tests announced by the company.
The release comes just two months after Opus 4.8 and follows a rapid rollout across Anthropic's 5-series lineup, including Mythos 5, Fable 5 and Sonnet 5. With this update, the lightweight Haiku model remains the only major version still awaiting a 5-series refresh.
Anthropic says Opus 5 is designed to be especially effective at self-checking and iterative problem-solving, improving its ability to refine answers until it reaches a workable result. In one example highlighted by the company, the model created a computer vision pipeline from an incomplete prompt.
The new model also arrives with fewer restrictions than Fable, while still keeping targeted safeguards in place for cybersecurity-related tasks. Anthropic notes that the system is built to limit risky uses while preserving legitimate defensive research and development workflows.
Another notable addition is Automatic Fallbacks, a beta feature that can redirect a request to a less powerful model when a safety filter is triggered. For API users, this means a response can still be delivered instead of a blocked request.
With Opus 5, Anthropic is signaling a broader shift toward AI systems that are more capable, more adaptable, and easier to use in real-world settings. This direction may shape the next phase of practical, responsible AI adoption.
