Tag
1 article
Anthropic claims its unreleased AI model outperforms Mythos 5.1 in both capability and alignment, successfully preventing malicious code injection in cybersecurity tests.