“`html
Anthropic has released Opus 5, the latest iteration of its flagship AI model. The new offering arrives just two months after the launch of Opus 4.8 and represents a significant addition to Anthropic’s expanding model lineup, which now includes versions across multiple capability tiers.
A key advantage of Opus 5 over the competing Fable model is its improved cost-efficiency and fewer operational restrictions. According to benchmark tests, Opus 5 actually surpasses Fable 5 in several performance metrics while remaining more affordable. The model demonstrates particular strength in self-verification and iterative problem-solving, as demonstrated by its ability to construct computer vision systems from incomplete instructions.
Privacy protections also distinguish Opus 5 from Fable, as it remains exempt from the 30-day data retention requirements applied to other models. Additionally, Anthropic has implemented a lighter regulatory touch on Opus 5, with safety classifiers expected to trigger 85 percent less frequently compared to Fable 5. The company is simultaneously introducing Automatic Fallbacks, a new feature allowing users to automatically route requests to less powerful models when safety protocols activate, ensuring functional responses rather than error messages for API users.
“`
Originally reported by Techcrunch →
About the Author
