Anthropic company released an updated version of the Opus model, which falls short in terms of cybersecurity applications compared to Mythos.

Anthropic PBC launches new version of its most powerful artificial intelligence model, Opus 4.7, just one week after releasing the more advanced Mythos. The company announced on Thursday that Opus 4.7 aims to improve software engineering capabilities, including handling difficult coding tasks that previously required more human supervision. However, the company stated that this new model still falls short of Mythos in overall capabilities, especially in terms of cybersecurity applications. According to the company's blog, during the training process of Opus 4.7, Anthropic even attempted to weaken the model's network defense capabilities through "differential weakening".