Claude Fable 5 is Anthropic's most powerful public model

Anthropic puts its most powerful model in public hands

On Tuesday 9 June 2026, Anthropic released Claude Fable 5. The company calls it the most capable model it has ever made available to the public.

Fable 5 is the first in a new tier Anthropic calls Mythos-class. This tier sits above the Opus models that powered Claude until now. The claim is simple. This is the strongest Claude most people can use.

The release raises a fair question. Is this a genuine step forward, or smart marketing in a year of steady, incremental updates? Both can be true at once. Our view is that for most UK businesses, the answer matters less than it first appears.

What actually changed

Anthropic first showed a Mythos model in April 2026. It held the model back from public release. The reason was safety. The model was unusually good at finding and exploiting software vulnerabilities.

That capability worried Anthropic enough to keep Mythos behind closed doors. It went only to vetted partners through a programme called Project Glasswing.

Fable 5 changes that. It is the same underlying model as the new Claude Mythos 5, with one difference. Fable 5 carries safety classifiers that block high-risk requests.

When a query touches cybersecurity, biology, chemistry, or distillation, Fable 5 redirects it. The request goes to the older Claude Opus 4.8 instead. Anthropic says these safeguards fire in under five percent of sessions. Its security team spent over 1,000 hours trying to break them and found no working route.

So the headline is not a smaller, safer model. It is the full model with a stronger cage around it. Mythos 5, with the cage partly removed, stays restricted to approved partners.

The two models even share a naming logic. Fable comes from the Latin fabula, meaning a thing that is told. That echoes the Greek mythos. The names point at the safeguards, not at any gap in raw ability.

A real leap, or clever positioning

Anthropic reports strong gains in software engineering, knowledge work, vision, and scientific research. It says the lead grows on longer and more complex tasks.

That pattern is worth noting. The benefits show up most on hard, multi-step work. They show up least on the short, everyday tasks that fill a typical working day.

The timing tells a story too. The launch lands days after Anthropic warned that frontier AI is becoming dangerously capable. It also arrives as the company prepares to enter public markets.

A more powerful model, framed around safety, released into a busy news cycle, is good positioning. That does not make the capability gains fake. It does mean you should read the announcement with a clear head.

For a developer shipping software all day, that lead could be real and useful. For an accounts team drafting client updates, it will barely register. The value of a frontier model depends on what you do with it.

This is the pattern across the current wave of releases. Each new model beats the last on paper. The real-world difference for a typical user is often smaller than the headline suggests. Only time and independent testing will show how much of this holds up in real work.

Why this matters for UK SMEs

Here is the part that gets missed. For most small and medium businesses, model power was never the limit.

We see this in nearly every engagement. The tools already on the desk can do far more than the team uses. The gap is rarely capability. It is confidence, habit, and clear ways of working.

At gecco we treat AI adoption as 80 percent people and culture, 20 percent technology. A stronger model does not change that ratio. A team that does not use Opus well will not suddenly use Fable 5 well.

A clear starting point matters more than a powerful model. That means understanding your processes, your data, and where time actually goes. Sort that out and almost any modern model performs well. Skip it and the most capable model in the world sits unused.

There is a cost reality too. Fable 5 is priced at roughly double the previous top Claude model. It is the most expensive mainstream model on the market. Most SME work does not need that much power.

Drafting an email, summarising a call, answering a customer query. These run well on cheaper models. Paying a premium for a frontier model to write a meeting note is poor value.

What you can do this week

The catch worth knowing

The rollout is staged. Fable 5 is free on Pro, Max, Team, and seat-based Enterprise plans until 22 June. From 23 June, Anthropic removes it from those plans until capacity grows. After that, using it needs paid usage credits.

Pricing also sits in US dollars and moves with usage. For a small team, that makes costs harder to predict than a flat monthly licence.

The safeguards add friction as well. Because the classifiers are tuned cautiously, they sometimes send harmless requests to the weaker fallback model. Anthropic acknowledges this and plans to narrow it.

None of this is a reason to ignore Fable 5. It is a reason to be deliberate. A more capable model is useful only when it meets a problem worth its cost.

Next step

The newest model is rarely the right place to start. Knowing where AI fits in your business is. If you want a clear read on where your team sits with AI adoption, take the AI readiness assessment.