Claude Fable 5: Mid-Tier Results on Coding Tasks
Claude Fable 5, the latest Mythos-class model from Anthropic, was tested on 200 real-world coding tasks. Despite high expectations, its performance remained average, scoring 59.8% in functionality and only 19.0% in security.