What Do You Need to Know About Claude Fable 5?

On June 9, 2026, Anthropic released Claude Fable 5, calling it the most capable model it has ever made available to the general public. Anthropic launched Claude Fable 5 on June 9, 2026, calling it the most capable model it has ever released to the general public. For security teams, CISOs, and anyone tracking the AI race, this launch is worth a close look. It is powerful, it is gated by new safety controls, and it arrives at a moment when Anthropic is openly warning that frontier AI is getting harder to control.

Here is what you actually need to know.

What Is Claude Fable 5?

Fable 5 is the first publicly available model in Anthropic’s new top tier, called the Mythos class. Mythos-class models are a tier of Claude models that sit above the Opus class in capability. In simple terms, this sits above Opus, which used to be Anthropic’s strongest line.

The name itself tells part of the story. Fable comes from the Latin “fabula,” meaning “that which is told,” which is close to the Greek word “mythos.” The safeguards are what distinguish the two models, Fable and Mythos, and are why they have different names.

So Fable 5 and Mythos 5 are the same underlying model. The difference is the cage built around it.

Fable 5 vs. Mythos 5: The Key Difference

This is the part most people get confused about, so let’s keep it clear.

Claude Mythos 5 is the same underlying model as Fable 5, but with the safeguards lifted in some areas. Mythos 5 is not for the public. It is being deployed through Project Glasswing, in collaboration with the US government, as an upgrade to Claude Mythos Preview, and it has the strongest cybersecurity capabilities of any model in the world.

Fable 5 is the version Anthropic judged safe enough to release to everyone. Mythos 5 is the restricted twin reserved for approved cyber defenders and infrastructure providers.

Why the Safety Controls Matter

Releasing a model this capable comes with risks, and without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage.

To handle this, Anthropic built a system of classifiers, which are separate AI systems that watch for misuse. When Fable’s classifiers detect a request related to cybersecurity, biology, chemistry, or distillation, the response is automatically handled by Claude Opus 4.8 instead, and users are informed whenever this occurs.

Fable 5 classifier’s deny message

In other words, if you ask Fable 5 something in a high-risk area, you do not get a flat refusal. The question quietly falls back to Opus 4.8, which is still a strong model. Allegedly, Anthropic’s early data shows that more than 95% of Fable sessions involve no fallback at all.

According to the firm, the controls are deliberately strict for now. They tuned the safeguards to be cautious, so they sometimes catch benign requests, and the company aims to reduce false positives as it refines the safeguards after launch.

The Three Areas Covered by Classifiers

1. Cybersecurity: Mythos-class models excel at discovering and exploiting software vulnerabilities, which can make cyberattacks easier and cheaper to commit. The classifiers are designed to block both exploitation and broader offensive cyber tasks like reconnaissance and lateral movement.
2. Biology and chemistry: Anthropic widened its biosecurity controls beyond a narrow set of bioweapons queries, citing concern about well-resourced actors and the models’ growing ability to handle real scientific tasks.
3. Distillation: Requests flagged as attempts to extract (“distill”) Fable 5’s capabilities to train competing models will fall back to Opus 4.8.

How Capable Is Fable 5, Really?

The headline claim is strong. Fable 5’s capabilities exceed those of any model Anthropic has made generally available, and it is state-of-the-art on nearly all tested benchmarks, with the largest lead on longer and more complex tasks.

A few examples that Claude itself promotes are:

• Software engineering: During early testing, Stripe reported that Fable 5 performed a codebase-wide migration in a 50-million-line Ruby codebase in a single day, work that would have taken a team over two months by hand.
• Long-running work: The model can work for days at a time in an agent harness like Claude Code, planning its approach, checking progress against the goal, and refining its work as it goes.
• Vision: Fable 5 can extract precise numbers from detailed scientific figures and rebuild a web app’s source code from screenshots alone.

The launch blog includes statistics and video examples of the capabilities.

A time-lapse of Claude playing Pokémon FireRed from start to finish using only raw game screenshots

Pricing and Availability

Both Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens.

For subscription users, the rollout is staged. 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 pull it from those plans, requiring usage credits going forward, with plans to restore it as a standard feature as soon as possible. On the API and consumption-based Enterprise plans, it is fully available now.

One detail worth flagging for compliance teams: Fable 5 and Mythos 5 are designated Covered Models, which carry 30-day data retention and are not available under zero data retention. The company says it will not use this data to train new models or for any non-safety purpose, and that it has added privacy protections, including logging all human access and deleting the data after 30 days in almost all cases.

Why the Timing Is Notable

The launch landed right after the firm raised an alarm about the broader direction of AI. Fable’s launch follows the company’s plea urging major global AI labs to establish a coordinated brake pedal on frontier AI development, and a warning that systems are advancing so rapidly they may soon achieve recursive self-improvement.

What This Means for Security Teams

For defenders, the practical takeaways are simple:

• Allegedly, the most powerful commercial AI capabilities are now in far more hands, which raises the baseline for what attackers might attempt.
• Mythos 5 stays locked to vetted defenders, so the offensive-defensive gap is being managed deliberately.
• Expect more AI-assisted activity on both sides, and plan your detection and monitoring accordingly.

The Bottom Line

Claude Fable 5 is the clearest sign yet that frontier AI capability and frontier AI risk are now moving together. The model is fast, capable, and built for long, complex work. The safety classifiers are the price of admission for letting the public use it. For security leaders, the smart move is to understand both halves of this story, because the same power that helps your team can also help the people you defend against.

See SOCRadar’s AI Threat Landscape Report: Hacker’s AI Advantage.