Summarized by Dodly:
Claude Fable 5: A Safer, Yet Limited, AI Release?
Sam Witteveen (Subscribed)
Audio Summary
Summary
Anthropic has launched Claude Fable 5, a rebranded version of its Mythos model, emphasizing safety for general use. While Anthropic claims Fable 5 surpasses previous generally available models, not all of Mythos 5's capabilities are included. Pricing for Fable 5 is double that of Opus 4.8, but less than half the price of the original Mythos preview, potentially due to expanded compute capacity. Benchmarks show Fable 5 performing significantly better than Opus 4.8 and even GPT 5.5 in some areas like coding, though not universally across all tasks. However, Fable 5 features significant safety restrictions, automatically flagging and switching to Opus 4.8 for topics like cybersecurity, malware, and even certain life science queries, suggesting a highly sensitive safety classifier. Anthropic is also implementing a 30-day data retention policy for Mythos class models, which could impact users concerned about data privacy on third-party platforms. Access to Fable 5 on subscription plans is currently limited, with plans to potentially reintroduce it to higher tiers.