The same model, two levels of safeguards: prices per token remain stable but hiding costs 75% less, and false safety positives fall sharply.
Anthropic presented Claude Fable 5.1 and Claude Mythos 5.1, two variations of the same model that differ only in their level of safeguards: Fable 5.1 is available for all, while Mythos 5.1 remains reserved for validated access programs, with protections designed for sensitive uses in cybersecurity and life sciences. Rates per token remain unchanged, but new models cost about 25% less for most customers, and the cost of setting the context down by 75%.
Anthropic also points to a sharp reduction in false positives of its security filters: the cybersecurity filter would trigger about 60% less often per session, and the biology filter 85% less often on simple medical issues. According to the company, Fable 5.1 exceeds Fable 5, Opus 5 and OpenAI GPT-5.6 Sol on several test benches, with a gap widening to the highest calculation effort levels.
Read the full article on the original source
Go to the source →