They named the restricted one Mythos. The public one Fable. One is the truth. The other is the story they are comfortable telling.
On June 9, Anthropic released Claude Fable 5, the first publicly available version of the Mythos model family it had been keeping behind closed doors since April 2026. The release comes with guardrails, fallbacks, and a clear message: the full version of this technology is still not something Anthropic is willing to put in everyone's hands.
What Mythos Was and Why It Stayed Locked
Claude Mythos Preview launched in April 2026 as the most capable AI model Anthropic had ever built. It was not released publicly. Instead, access was restricted to a small group of trusted organizations through a program called Project Glasswing, with a focus on critical infrastructure providers and cybersecurity researchers. The reason was straightforward. Mythos had demonstrated the ability to autonomously discover over 10,000 critical zero-day vulnerabilities across every major operating system and browser during its first month of operation. It could develop professional-grade cyberattacks without human assistance. Anthropic's position was that releasing a model with those capabilities to the general public, without serious controls in place, was not responsible. Last week, Anthropic expanded Mythos access to hundreds of organizations across 15 countries, still focused on critical infrastructure. Then, on June 9, the public version arrived.What Fable 5 Actually Is
Fable 5 is not a watered-down Mythos. It is the same underlying model with a layer of hard safety controls applied on top. Anthropic describes it as a Mythos-class model made safe for general use, and its benchmark performance supports that framing. It leads on nearly every tested capability metric, with performance advantages that grow as tasks get longer and more complex. The practical difference between Fable 5 and Mythos 5 is what happens when a query touches certain domains. Requests related to cybersecurity, biology and chemistry, and model distillation are automatically routed to Claude Opus 4.8, the next-most-capable model, instead of being answered by Fable directly. Users are notified when this happens. Anthropic says this fallback triggers in less than 5% of sessions on average. That 5% figure matters. It means the vast majority of what Fable 5 does, it does without restriction. The guardrails are targeted at a narrow but critical set of use cases where the capability gap between Fable and everything else creates genuine risk.The Part That Deserves More Attention
Anthropic released Fable 5 the same week it published a public letter urging major AI labs to establish a coordinated brake on frontier AI development. The letter warned that AI systems are advancing fast enough that recursive self-improvement, where models autonomously improve themselves without human intervention, may be approaching. Read that sequence again. Anthropic warned the world that AI might be getting too powerful to control, then released its most powerful public model ever the same week. This is not hypocrisy. It is the tension that defines every serious AI lab right now. If you believe the technology is coming regardless of what any single organization does, the strategic choice is to be the one releasing it with the most careful controls rather than ceding that ground to competitors who might not bother. Fable 5 is Anthropic's version of that bet. Whether the guardrails are sufficient is a different question. The company says more than 1,000 hours of red-teaming found no universal jailbreaks. That is a meaningful bar. It is not the same as saying the safeguards will hold under adversarial pressure at scale from millions of users.Read also: Morgan Stanley Just Let Clients Lend Their Bitcoin to Wall Street. That Should Make You Think.
What This Means for Crypto and AI Intersection
Two things connect directly to what BitBrainers covers. First, Anthropic is preparing to go public. The Fable 5 launch happens as the company approaches its IPO, alongside OpenAI. A model release of this magnitude, with this level of public attention, is also a valuation story. Mythos-class capability at a consumer price point changes Anthropic's addressable market dramatically. Second, the cybersecurity fallback has direct implications for crypto. Blockchain security, smart contract auditing, and protocol vulnerability research all sit in the category of domains where Fable 5 routes queries to Opus 4.8. That is not necessarily a problem for legitimate research, but it does mean that the most capable AI tool currently available to the public is deliberately less capable in the exact domain where crypto protocols have the most exposure. The 10,000 zero-day vulnerabilities Mythos found across mainstream operating systems are a preview of what a model this capable could do if pointed at on-chain code at scale.
Anthropic perpetual futures | Source: TradingView
Availability and What Happens After June 22
Fable 5 is included at no extra cost on Pro, Max, Team, and seat-based Enterprise plans through June 22. On June 23, it moves behind usage credits. Anthropic has said it aims to restore Fable 5 as a standard subscription feature as quickly as capacity allows, but gave no firm timeline. The API pricing is $10 per million input tokens and $50 per million output tokens, which Anthropic notes is less than half the cost of the original Mythos Preview. For developers building on the API, Fable 5 is available immediately with no staged rollout. Mythos 5, the version with fewer restrictions, remains limited to organizations already approved through Project Glasswing and will expand through a broader trusted access program over time.On The Radar
Watch how Fable 5 performs on real-world tasks over the next two weeks while it is still free on subscription plans. Watch whether any meaningful jailbreaks emerge and how quickly Anthropic patches them. That will tell you more about the robustness of the guardrails than any press release. Watch the Anthropic IPO timeline, because Fable 5's reception is part of that story. And watch how OpenAI responds, because a capability release of this magnitude never sits in a vacuum for long.Sources
Anthropic | TechCrunch | VentureBeatBitBrainers. We check the facts so you don't have to.
Disclosure: This article is for informational purposes only and does not constitute financial advice. BitBrainers may earn affiliate commissions through links on this site.
— BitBrainers Editorial
