Anthropic debuts Claude Fable 5, safer Mythos model with safeguards
Anthropic just rolled out Claude Fable 5, a new and safer version of its Mythos language model.
While the original Mythos was only available to big names like Amazon Web Services and Google, Fable 5 is designed for a wider crowd and comes with extra safety features to prevent misuse, especially in sensitive areas like cybersecurity and biology.
Fable 5 switches to restricted mode
Fable 5 has built-in safeguards that kick in during risky situations, automatically switching to a more restricted mode if needed.
Testing shows that at least 95% of sessions run smoothly on their own. Right now, customer companies like Base44 and Genspark are trying it out for things like app development and UI design.
If you were on Pro, Max, Team, or seat-based Enterprise plans, Fable 5 was free through June 22, 2026 (the free access window ran June 10-22, 2026); starting June 23, 2026 it required usage credits.
After that, it cost $10 per million input tokens and $50 per million output tokens (which is about twice the price of Claude Opus 4.8).
Anthropic says it plans to bring it back into regular subscription plans soon.