Anthropic’s Claude Fable 5 is the version of the Mythos that is publicly accessible today

Anthropic is bringing its most powerful AI model to the general public for the first time, but it’s doing so with guardrails.

On Tuesday, the AI ​​company Claude launched Fable 5, the first publicly available version of its Mythos model. Anthropic says Fable 5 excels at software engineering, cognitive work, and vision, but comes with strict security limits. In high-risk fields such as cybersecurity, biology, chemistry, and distillationthe form blocks responses and reverts to Claude Opus 4.8.

Mythos launched as a preview in April, and was initially limited to a few partners due to cybersecurity concerns. Last week Anthropy Expand your reach to hundreds of organizations Across 15 countries, again focusing on organizations that manage critical infrastructure.

Now the release of this technology is available to anyone through Anthropic’s Claude API and consumption-based enterprise plans. Access to subscriptions will be rolled out in phases: Through June 22, Fable 5 will be included in seat-based Pro, Max, Team, and Enterprise plans at no additional cost. On June 23, Anthropic will pull Fable 5 from those plans, requiring usage credits going forward, with plans to bring it back as a standard subscription feature as soon as possible.

Anthropic is also publishing a new version of Mythos, called Mythos 5, for institutions that have already been approved to access the advanced model.

The launch of Fable comes as Anthropic prepares to enter the public markets alongside OpenAI And Elon Musk SpaceX. It also follows Artificial Intelligence Company Call Urge major global AI labs to create a coordinated brake pedal for frontier AI development. Anthropic has warned that systems are advancing so rapidly that they may soon achieve recursive self-improvement (RSI), where they improve themselves autonomously without human intervention.

Concerned about what a Mythos-class model could do if it fell into the wrong hands, Anthropic says it tested its classifiers with jailbreak attempts before Fable 5 was released.

“Internally, we implemented an external bug bounty, which produced no global jailbreaks in over 1,000 hours of testing. We then worked with external red team organizations that also failed to find global jailbreaks.”

However, it is possible that there will be new attacks. As a result, with the launch of Fable 5 and Mythos 5, Anthropic said it would require 30-day retention on all traffic, even if companies had previously entered into no-data retention agreements. The company said it will not use the data for training and will only use it to “defend against sophisticated and new attacks, including new jailbreaks,” and “identify and reduce false positives.” This policy could set a precedent in the industry, where access to robust models increasingly comes with mandatory data retention policies as part of safety measures.

For those who continue to use the form, not every question will get a Fable 5 answer. Anthropic says instances where Fable has to resort to Opus 4.8 are rare, as early data shows that at least 95% of Fable sessions run entirely on the form’s own responses.

In third-party testing, analytics firm Hex said in a statement that Fable was first to score 90% on its core analytics benchmark for complex, long-term analytical tasks.

“On the most difficult questions, he shows strong judgment and attention to nuance,” Hicks said.

Vibe coding platform Base44 noted in a statement that Fable is better at “full, one-shot implementations” and has excellent tool calling ability. AI-powered workspace and agent platform Genspark said Fable outperformed all other models in its evaluations and performed much better on tasks like UI design and game coding.

Both Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, which is double the price of Opus 4.8. This price alone may act as a deterrent to widespread use.

Many companies have since become critical of the costs of AI See the bills coming Or spend their annual AI budgets early. Advanced models such as Opus 4.8 can exacerbate these problems, with advanced reasoning skills that can break down a single request into multiple tasks.

Anthropic said it expects demand for Fable 5 to be very high and difficult to predict. In fact, some, like shopping rewards platform Rakuten, may think the upside is worth the price point.

“With the best effort, Fable reflects and validates its work,” Rakuten said in a statement. “For us, that’s what makes highly autonomous operations possible. The extra thinking pays for itself.”