What Anthropic released
On June 9, 2026, Anthropic launched Claude Fable 5: a Mythos-class model made safe for general use. Fable 5's capabilities exceed those of any model Anthropic has ever made generally available. It is state-of-the-art on nearly all tested benchmarks of AI capability, showing exceptional performance in software engineering, knowledge work, vision, scientific research, and many other areas. The longer and more complex the task, the larger Fable 5's lead over every other Anthropic model. CoSN
Alongside it, Anthropic released Claude Mythos 5 for a small group of cybersecurity defenders and infrastructure providers — the same underlying model as Fable 5, but with some safeguards lifted. The shared root is deliberate: "Fable," from the Latin fabula, "that which is told," is akin to the Greek mythos. What separates the two is not the model but the guardrails around it. K-12 Dive
Available immediately on Claude API, Claude Platform, Claude Code, consumption-based Enterprise plans, and on AWS, Google Cloud, and Microsoft Foundry. Pricing is $10 per million input tokens and $50 per million output tokens — exactly double Opus 4.8, but 90% less with prompt caching. On benchmarks, Fable 5 scores 80.3% on SWE-Bench Pro compared to Opus 4.8's 69.2% and GPT-5.5's score. The models carry a 1 million token context window and 128,000 maximum output tokens, with a knowledge cutoff of January 2026. EdWeekK-12 Dive
The backstory — why this release took two months
To understand why Fable 5 matters, you have to go back to April 7, 2026 — the day AI Ready School covered what was then the most unusual product announcement in AI history.
Launched as a preview in April, Mythos was initially limited to a handful of partners due to cybersecurity concerns. Last week, Anthropic expanded access to hundreds of organisations across 15 countries, focusing on organisations that manage critical infrastructure. Now a version of that technology is available to anyone through Anthropic's Claude API and consumption-based Enterprise plans. eSchool News
Two months after Anthropic rolled out Mythos to a limited number of users, citing concerns about the AI model's potential to do damage in the wrong hands, the company said it is ready to release an equally powerful model to the public. The broad release is possible because of new safeguards that block responses in specific high-risk areas, including cybersecurity and biology. aireadyschool
Wary of what a Mythos-class model could do in the wrong hands, Anthropic stress-tested its classifiers. Releasing a model this capable comes with risks. Without safeguards, Fable 5's capabilities in areas like cybersecurity could be misused to cause serious damage. The solution Anthropic arrived at is architecturally elegant and historically significant: build the full model, apply hard safety classifiers to the highest-risk capability domains, and release the result to the world. The dangerous capabilities are not removed. They are gated. The intelligence is identical. What changes is the passport it travels on. eSchool NewsCoSN
How the safety architecture actually works
In high-risk areas like cybersecurity, biology, chemistry, and distillation, the model blocks responses and falls back to Claude Opus 4.8. Anthropic made Fable 5 available across all of their surfaces — the Claude.ai chat interface, Claude Code for web, Claude Code CLI, and Claude Cowork. eSchool NewsK-12 Dive
Anthropic's platform documentation states that refusals return a stop reason of "refusal" as an HTTP 200 and are not billed if no output is generated. In practice, this means that fewer than 5% of sessions encounter the classifier. For the other 95% — the software engineers, researchers, writers, teachers, students, and knowledge workers who will use Fable 5 every day — the full power of a Mythos-class model is available, without restriction, for the first time. Discovery Education
This is the most significant real-world deployment of Constitutional AI principles ever attempted at scale. Not as a research paper. Not as a controlled pilot. As a live product used by millions of people, where the safety layer is not a human reviewer or a policy document but a classifier woven into the model's own reasoning. Whether it holds — whether the 5% of dangerous requests are reliably contained across the full distribution of users, use cases, and adversarial attempts — is the question the next six months will answer. Anthropic's IPO, expected as soon as this year, depends significantly on the answer.
What Fable 5 can actually do — for the problems that matter
"The best way to describe Fable is that it feels big. Not just in terms of speed and cost, but also in how much it knows." That description, from one of the first detailed independent evaluations published within hours of release, captures something the benchmarks alone do not. Fable 5 is not just faster or more accurate on the tasks previous models handled. It handles a different class of task entirely. K-12 Dive
The longer and more complex the task, the larger Fable 5's lead over other models. This is the most important sentence in Anthropic's entire release announcement. It means that Fable 5's advantage is not at the easy end of the distribution — the questions that any competent model can answer. It is at the hard end: multi-step reasoning across vast contexts, ambiguous problems with no single correct answer, scientific research that requires synthesising hundreds of sources, software engineering tasks that span entire codebases, knowledge work that demands genuine understanding rather than pattern matching. CoSN
These are exactly the tasks that define high-value human work. And a model that leads on them — available to anyone, at $10 per million input tokens — changes the economics of every knowledge-based profession on earth. Not in the future. Starting yesterday.
What this moment means — read alongside everything that has come before
AI Ready School has been tracking global AI developments every week for two months. In that time, your students have been growing up alongside DeepSeek's open-source disruption, GPT-5.5's agentic leap, Snap's 1,000 layoffs, the Harvard study showing AI outdiagnosing doctors, and the creativity research showing AI matching average human originality. Each story has pointed in the same direction: AI capability is advancing faster than institutional understanding of it, and the gap between what AI can do and what schools are preparing children for is growing, not closing.
Fable 5 is not a new direction. It is the same direction, accelerated.
The launch comes as Anthropic prepares to enter the public markets, alongside OpenAI and Elon Musk's SpaceX. It also follows the AI firm's plea urging major global AI labs to establish a coordinated brake pedal on frontier AI development. Anthropic warned that systems are advancing so rapidly that they may soon achieve recursive self-improvement — autonomously improving themselves without human intervention. eSchool News
Read that carefully. The company that just released its most powerful model to the public also issued a warning, in the same week, that AI may soon improve itself without human intervention. This is not contradiction. It is the precise tension that defines the moment we are in — and the moment every school must now navigate. The technology is arriving whether institutions are ready or not. The only choice available is whether to engage with it deliberately, with values and a philosophy intact, or to be overtaken by it.
Cypher — which asks students to think rather than simply receive answers — was built for a world where AI answers are free and abundant. In a world where Fable 5 is available to any student on any device, the question of what school is actually for becomes urgent in a new way. Not "can you find the answer?" AI can. "Can you ask the right question?" AI cannot — not yet, not reliably, not in the way that a child who has been trained to push past the obvious can. The distinction between the student who uses Fable 5 to do their thinking and the student who uses it to sharpen their thinking is the distinction that will define the next generation of professionals, researchers, and leaders.
Matrix — AI Ready School's on-premise infrastructure — now sits in a world where Mythos-class intelligence is publicly available. The schools that have built local AI infrastructure are not just protected from data sovereignty concerns. They are positioned to evaluate, deploy, and govern frontier-class models within their own walls — on their own terms, with their own safeguards, accountable to their own communities. The question of which AI a school uses, and how it is governed, is no longer a procurement decision. It is an ethical and educational one.
The line from Anthropic's own announcement
"Today we're launching Claude Fable 5: a Mythos-class model that we've made safe for general use." CoSN
Two months ago, Mythos was locked away. Yesterday, it was made available to anyone with an internet connection. That is the speed at which the frontier is moving. The schools that read this update, understand what it means, and act accordingly are the ones that will be genuinely ready for the world their students are graduating into.
