Anthropic Just Shipped a Model It Once Called Too Risky
Anthropic released Claude Fable 5 on June 9, a public build of its restricted Mythos model wrapped in new safeguards.
Key Takeaways
- Fable 5 is the first public version of Anthropic's Mythos system, wrapped in safeguards that redirect sensitive queries to Opus 4.8.
- Anthropic is splitting access into a guarded public model and a less restricted Mythos 5 for vetted cyberdefenders, signaling a new release strategy.
- Fable 5 ships free on paid plans only through June 22, after which use requires usage credits.
Anthropic released Claude Fable 5 on Tuesday (June 9), its first public model built on the closely watched Mythos system. The company says the model is safe for general use while still ranking among its most capable releases yet.
The move matters because Mythos was once held back as too powerful for open release. Anthropic is now testing whether a guarded public version can ship without exposing its riskiest capabilities.
According to Anthropic, Fable 5 outperforms its Opus models on longer and more complex tasks. The company says the model can also work autonomously for longer than any prior Claude release.
Anthropic says Fable demonstrated exceptional performance across software engineering, knowledge work, vision, and scientific research. The model is being released with conservative safeguards to prevent misuse in areas like cybersecurity.
According to Anthropic, sensitive queries are redirected to Opus 4.8 instead of Fable 5. Most requests touching cybersecurity, chemistry, and biology will get answers from the older model.
The company says these safeguards are expected to trigger in less than 5% of sessions on average. Tom's Guide reports the system works in at least 95% of sessions without falling back to another model.
A split between public and trusted access
Anthropic is also releasing Claude Mythos 5 for a small group of cyberdefenders and infrastructure providers. It uses the same underlying model as Fable 5, but with some safeguards lifted.
According to Anthropic, Mythos 5 has the strongest cybersecurity capabilities of any model in the world. It is deployed through Project Glasswing as an upgrade to the earlier Claude Mythos Preview.
Tom's Guide reports that Mythos was first limited to organizations through Project Glasswing in April. That program was built with the U.S. government and focused on cyberdefense and critical infrastructure.
According to MacRumors, Mythos 5 access includes Apple, one of Anthropic's Project Glasswing partners. Access is set to expand through a broader trusted access program.
Pricing and how to try it
According to MacRumors, Fable 5 and Mythos 5 cost $10 per million input tokens and $50 per million output tokens. That price is less than half the cost of the earlier Claude Mythos Preview.
MacRumors reports that Fable 5 is included in Pro, Max, Team, and seat-based Enterprise plans through June 22. On June 23 the model leaves those plans, and further use will require usage credits.
Early testing suggests the model handles ambitious projects from a single prompt. TechCrunch reports that researcher Ethan Mollick used Fable to generate several playable video games through one initial prompt in Claude Code.
According to TechCrunch, Mollick said Fable outperformed basically every other public model he had used by a considerable margin. He also said it would work up to a dozen hours executing on multi-page specifications.
About the author
The Mixstackrr Team is a group of writers and editors with more than 10 years of combined experience in SEO and consumer tech. We test devices, dig through settings, and turn everyday tech problems into clear, step-by-step guides anyone can follow.
