Anthropic debuts Claude Fable 5, a public version of Mythos

Anthropic is making its most powerful AI model available to the general public for the first time — but with safety measures in place.

On Tuesday, the company launched Claude Fable 5, the first public release of its Mythos model. According to Anthropic, Fable 5 performs well in software engineering, knowledge work, and vision tasks, but includes strict safety restrictions. In high-risk domains such as cybersecurity, biology, chemistry, and distillation, the model will refuse to answer and instead defer to Claude Opus 4.8.

Initially released as a preview in April, Mythos was only available to a small number of partners because of cybersecurity concerns. Last week, Anthropic extended access to hundreds of organizations in 15 countries, once again prioritizing those that manage critical infrastructure.

Now, a version of that technology is accessible to everyone via Anthropic's Claude API and consumption-based Enterprise plans. Subscription access will be rolled out in phases: until June 22, Fable 5 is included in Pro, Max, Team, and seat-based Enterprise plans at no additional charge. On June 23, Anthropic will remove Fable 5 from those plans, requiring usage credits thereafter, with the intention of reinstating it as a standard subscription feature as soon as possible.

Anthropic is also rolling out a new version of Mythos, named Mythos 5, to organizations already approved for access to the advanced model.

Fable's launch coincides with Anthropic's preparations to enter public markets, alongside OpenAI and Elon Musk's SpaceX. It also follows the company's call for major global AI labs to create a coordinated "brake pedal" for frontier AI development. Anthropic warned that AI systems are advancing so quickly that they may soon achieve recursive self-improvement (RSI), enabling them to autonomously improve without human input.

Concerned about what a Mythos-class model could do in the wrong hands, Anthropic says it stress-tested its classifiers using jailbreak attempts before releasing Fable 5.

“Internally, we ran an external bug bounty program that found no universal jailbreaks in over 1,000 hours of testing. We then worked with external red-teaming organizations, which also failed to identify any universal jailbreaks.”

That said, novel attacks are still possible. As a result, with the launch of Fable 5 and Mythos 5, Anthropic said it will require a 30-day retention period for all traffic, even if enterprises previously had zero-retention agreements. The company stated it will not use the data for training, only to “defend against complex and novel attacks, including new jailbreaks” and to “identify and reduce false positives.” This policy could set an industry precedent, where access to increasingly powerful models comes with mandatory data-retention policies framed as a safety measure.

For those continuing to use the model, not every question will receive a Fable 5 answer. Anthropic says the instances where Fable must defer to Opus 4.8 are rare, with early data showing that at least 95% of Fable sessions rely entirely on the model's own responses.

In third-party testing, analytics company Hex stated that Fable was the first to achieve a 90% score on its core analytics benchmark, which involves complex, long-running analytical tasks.

“On the hardest questions, it demonstrates strong judgment and attentiveness to nuance,” Hex said.

Vibe-coding platform Base44 noted in a statement that Fable is better at “one-shotting full apps” and has excellent tool-calling capabilities. AI-powered workspace and agent platform Genspark said Fable outperformed every other model in its evaluations and performed significantly better on tasks such as UI design and game coding.

Pricing for both Fable 5 and Mythos 5 is set at $10 per million input tokens and $50 per million output tokens, double the cost of Opus 4.8. That price alone may discourage widespread adoption.

Many enterprises are becoming increasingly critical of AI costs after seeing their bills or exhausting their annual AI budgets early. Advanced models like Opus 4.8 can worsen these issues, as their sophisticated reasoning capabilities can break a single request into multiple tasks.

Anthropic said it expects demand for Fable 5 to be very high and hard to predict. Indeed, some companies, such as shopping rewards platform Rakuten, may consider the benefits worth the cost.

“At the highest effort level, Fable reflects on and validates its own work,” Rakuten said in a statement. “For us, that’s what makes highly autonomous operations possible — the extra thinking pays for itself.”