Anthropic brings its most powerful AI model to the public — with hard safety limits

Anthropic is making its most advanced AI model accessible to the general public for the first time, but with significant safety restrictions. On Tuesday, the company launched Claude Fable 5, a publicly available version of its Mythos model, which had previously been limited to a small number of vetted partners due to cybersecurity concerns.

What Fable 5 can — and cannot — do

Fable 5 is designed to excel at software engineering, complex knowledge work, and vision-based tasks. In third-party testing, analytics company Hex reported that Fable was the first model to achieve a 90% score on its core analytics benchmark for complex, long-running analytical tasks. AI-powered workspace platform Genspark noted that Fable outperformed every other model in its evaluations, particularly on UI design and game coding.

However, the model comes with hard safety limits. In high-risk areas such as cybersecurity, biology, chemistry, and model distillation, Fable 5 blocks responses and falls back to Anthropic’s earlier Claude Opus 4.8. The company says these deferrals are rare, with early data showing at least 95% of Fable sessions running entirely on the model’s own responses.

A cautious rollout after limited preview

Mythos was initially launched as a preview in April but was restricted to a handful of partners due to concerns about misuse. Last week, Anthropic expanded access to hundreds of organizations across 15 countries, focusing on those managing critical infrastructure. Now, a version of that technology is available to anyone through Anthropic’s Claude API and consumption-based Enterprise plans.

For subscription users, access will roll out in stages. Through June 22, Fable 5 is included in Pro, Max, Team, and seat-based Enterprise plans at no extra cost. On June 23, Anthropic will remove Fable 5 from those plans, requiring usage credits going forward, with plans to restore it as a standard subscription feature as soon as possible.

Data retention as a safety measure

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

Anthropic says it stress-tested its classifiers with jailbreak attempts before releasing Fable 5. Internally, an external bug bounty produced no universal jailbreaks in over 1,000 hours of testing. The company also worked with external red-teaming organizations, which also failed to find universal jailbreaks. However, Anthropic acknowledged that novel attacks may still be possible.

Why this matters for the AI industry

Fable’s launch comes as Anthropic prepares to enter the public markets, alongside OpenAI and Elon Musk’s SpaceX. It also follows the AI firm’s public plea urging major global AI labs to establish a coordinated “brake pedal” on frontier AI development. Anthropic has warned that systems are advancing so rapidly that they may soon achieve recursive self-improvement (RSI), autonomously improving themselves without human intervention.

Pricing for both Fable 5 and Mythos 5 is set at $10 per million input tokens and $50 per million output tokens — double the price of Opus 4.8. That cost alone may serve as a deterrent for widespread use, especially as many enterprises grow critical of AI costs after seeing bills climb or blowing through yearly AI budgets early. Advanced models like Opus 4.8 can exacerbate those issues, with reasoning skills that can split a single request into multiple tasks.

Still, some organizations see the value. Rakuten, the shopping rewards platform, noted in a statement: “At the highest effort, Fable reflects on and validates its own work. For us, that’s what makes highly autonomous operations possible — the extra thinking pays for itself.”

Conclusion

Anthropic’s launch of Claude Fable 5 represents a significant step in bringing frontier AI capabilities to a broader audience, but it also underscores the growing tension between innovation and safety. With mandatory data retention policies, hard safety limits, and high pricing, the company is signaling that powerful AI access will come with strings attached — a model that may shape how the industry approaches responsible deployment in the years ahead.

FAQs

Q1: What is the difference between Claude Fable 5 and Mythos 5?
Mythos 5 is Anthropic’s most advanced AI model, previously limited to approved partners. Claude Fable 5 is a publicly available version of Mythos 5, with additional safety guardrails that block responses in high-risk areas like cybersecurity and biology.

Q2: Will my data be used for training if I use Fable 5?
No. Anthropic requires a 30-day retention on all traffic but says it will not use the data for training. The retention is only for defending against novel attacks and reducing false positives.

Q3: How much does Fable 5 cost?
Pricing is $10 per million input tokens and $50 per million output tokens — double the cost of Anthropic’s previous model, Opus 4.8. Through June 22, it is included in subscription plans at no extra cost.