What Is Claude Fable 5?

On June 9, 2026, Anthropic released Claude Fable 5, the most capable version of Claude that anyone can use. It sits above Claude Opus, the model that until now was the top of the everyday lineup. Anthropic calls this new tier “Mythos-class,” and Fable 5 is the first one it has opened to the public. The short version for most people: it is a serious jump in capability aimed at very hard problems, you can probably try it free for a couple of weeks, and you almost certainly do not need it for daily tasks. Here is the plain-English tour.

What is Claude Fable 5?

Claude Fable 5 is a new AI model from Anthropic, the company behind Claude. Anthropic describes it as the “next generation of intelligence for the hardest knowledge work and coding problems.” In its own testing it ranks at or near the top on most major industry benchmarks, including coding, finance, and reading detailed scientific charts.

The name to understand here is Mythos. For months, Anthropic had a model class called Mythos that was more powerful than Opus but kept locked away, available only to a small number of vetted enterprise partners under a program called Project Glasswing. Fable 5 is the version of that powerful model that Anthropic decided was safe enough to release to everyone, with guardrails built in. So when you read “Mythos-class,” think: the tier above Opus that used to be off-limits, now opened up in a careful form.

Where does Fable fit in the Claude lineup?

Anthropic offers a ladder of models, from quick-and-cheap to slow-and-powerful. The bigger ones cost more and think harder. Fable 5 now sits at the very top. If you are new to all of this, our guide to the AI models of 2026 walks through the whole field, and Claude’s Free, Pro, and Max plans explains which subscription gives you what.

Model	Think of it as	Best for
Claude Haiku	The quick one	Fast, low-cost answers and high-volume simple tasks
Claude Sonnet	The everyday default	Most writing, research, and daily work. The balanced pick.
Claude Opus 4.8	The heavy lifter	Harder reasoning, longer projects, trickier problems
Claude Fable 5	The new frontier tier	The hardest coding and knowledge work, long multi-step jobs

The useful mental model: Sonnet is the car you drive every day. Opus is the truck you rent when you are moving house. Fable is the crane you call in when you are building a tower. Most people spend almost all their time in the car.

What can Fable 5 actually do?

Benchmarks are hard to feel. Here are the claims Anthropic published, translated into things you can picture:

• Work that used to take a team months, done in a day. The payments company Stripe reported that Fable 5 “compressed months of engineering into days.” In one case it updated a 50-million-line codebase in a single day, a job Anthropic says would have taken a team of engineers more than two months by hand.
• It can read pictures, not just words. Anthropic says Fable can pull exact numbers out of detailed scientific charts, and even rebuild a working app just from looking at screenshots of it. This matters for everyday tasks like getting precise figures out of a graph in a PDF.
• It stays on track for a very long time. Older models tend to lose the thread on long, multi-step jobs. Anthropic says Fable “stays focused across millions of tokens,” which is the unit AI uses to measure text. In a playful test, it finished a notoriously hard video game three times more often than Opus did.

None of this means it never makes mistakes. It means the ceiling is higher than anything Anthropic has put in public hands before. The “tokens” idea comes up a lot in AI pricing and limits, so if it is new to you, our plain-English explainer on tokens is worth five minutes.

Why did Anthropic warn about AI, then release this?

This is the part that made headlines, and it is worth understanding clearly. Just days before this launch, Anthropic publicly urged AI companies to agree on shared safety rules, warning that systems are advancing so fast they may soon reach “recursive self-improvement,” meaning AI that can improve itself without a human in the loop. Then the same company released its most powerful public model yet. Outlets like TechCrunch and CNBC noted the tension right away.

Anthropic’s answer is in how Fable is built. Rather than ship the raw Mythos model, it wrapped it in guardrails. The most important one: when you ask Fable something in a high-risk area, mainly offensive cybersecurity or detailed biology and chemistry, it does not answer with its full power. It quietly hands the request to the smaller Claude Opus 4.8 model instead, and you are not charged the Fable rate for it. Anthropic says more than 95% of sessions never trigger this fallback at all. It also says the model went through more than 1,000 hours of attempts to break its safety rules, plus outside red-team testing that found no universal way around them.

This design fits how we think about AI at Beginners in AI. The strongest setups keep a person in charge and use the powerful tool for the narrow job it is best at, while routing the dangerous edges somewhere safer. Fable handing risky questions back to a smaller, more constrained model is that idea built into the product.

Can I use Claude Fable 5 right now?

Most likely yes. Fable 5 is available in the Claude apps and on claude.ai, through the Claude developer platform, and on the big cloud platforms (Amazon Web Services, Google Cloud, and Microsoft Foundry). For regular users, the key detail is the calendar:

• June 9 to June 22, 2026: Fable 5 is included at no extra cost on the Pro, Max, Team, and Enterprise plans. This is the window to play with it for free.
• June 23, 2026 onward: using Fable will require pay-per-use credits. Anthropic says it hopes to fold Fable back into the standard plans once it has enough computing capacity.

If you are on a paid Claude plan, the practical move is to open the model picker in the next two weeks, switch to Fable, and try it on a real task you find hard. After June 22, decide whether the harder jobs in your week are worth paying per use. Not sure where to even begin with Claude? Our Claude beginner path starts from zero.

What does Claude Fable 5 cost?

Outside the free window, Fable is priced per million tokens, the standard way AI usage is billed. A million tokens is roughly 750,000 words, so the numbers are smaller in practice than they look.

What you pay for	Price
Input (the text you send in)	$10 per million tokens
Output (the text it writes back)	$50 per million tokens
Prompt caching	Up to 90% off repeated input

Anthropic says this is less than half the price of the earlier Mythos preview, and processing run only in the United States costs about 10% more. For everyday users on a subscription, none of this matters until the free window ends. For anyone building on the developer platform, the pricing is the thing to watch.

Do beginners actually need Fable 5?

For almost everyone reading this, the answer is no, and that is fine. Claude Sonnet handles the large majority of real work people do: drafting emails, summarizing documents, planning, answering questions, brainstorming. Opus covers the harder end. Fable is built for problems most of us rarely face, like rewriting a giant software project or running a multi-day research task on its own.

Use the most capable model that fits the job, not the most powerful one on the shelf.

Reaching for the biggest model on every task is a common beginner trap. It is slower, it can cost more, and it rarely produces a better email. The smarter habit is to match the tool to the work. When you hit a wall that Sonnet or Opus cannot get over, a long, tangled, multi-step job, that is the moment to bring Fable in. The rest of the time, the everyday models are not a compromise. They are the right choice.

How is Fable different from Mythos?

They are the same underlying model with different locks. Claude Fable 5 is the public version, with safety guardrails switched on and the risky-domain routing described above. Claude Mythos 5 is the same model with some of those guardrails lifted, available only to authorized users that Anthropic has vetted, the way Mythos has worked since it started under Project Glasswing. Think of Fable as the model with the safety catch on, and Mythos as the version a small set of approved partners can use with the catch released. For the public, Fable is the one that matters.

Common questions

Is Claude Fable 5 free?

For a short window, yes. From June 9 to June 22, 2026, Fable 5 is included at no extra cost on the Pro, Max, Team, and Enterprise plans. From June 23 onward it requires pay-per-use credits, though Anthropic says it hopes to bring it back into the standard plans later.

Is Fable better than Opus or Sonnet?

On raw capability, yes, Anthropic places Fable above both. But better on benchmarks is not the same as better for your task. For everyday writing and research, Sonnet is faster and cheaper and gives you the same quality of answer. Fable pulls ahead on the hardest, longest, most complex jobs.

What does Mythos-class mean?

It is Anthropic’s name for a capability tier that sits above its Opus models. Mythos-class models were kept restricted at first because of their power. Fable 5 is the first one released to the general public, with safety guardrails added.

Does Claude Fable keep my data?

Yes. Anthropic keeps 30 days of all Fable traffic for safety monitoring, even for customers who previously had zero-retention agreements. The company says this data is not used to train new models. As a rule, avoid pasting passwords or sensitive secrets into any AI model.

Should I switch to Fable for everyday tasks?

No need. For email, summaries, planning, and questions, Sonnet or Opus is the better fit. Try Fable during the free window on something hard, then keep using the everyday models for everyday work.

Sources

• Anthropic: Claude Fable 5 and Claude Mythos 5 (official announcement)
• Anthropic: Claude Fable product page
• TechCrunch: Anthropic releases Claude Fable 5 days after warning AI is getting too dangerous
• CNBC: Anthropic releases Mythos-like AI model to the public, Claude Fable 5
• AWS: Claude Fable 5 on Amazon Bedrock

You may also like

• What’s New in Claude in 2026
• The AI Models of 2026, Compared
• How to Use Claude AI
• Claude Free vs Pro vs Max
• The Claude Beginner Path
• The AI Glossary
• The daily Beginners in AI newsletter