FINALLY OFFLINE

CLAUDE FABLE 5 IS FREE UNTIL JULY 12. THE STORY OF HOW IT KEEPS DISAPPEARING

By Chief Editor | 7/7/2026

Claude Fable 5 is Anthropic's most capable model made generally available, a safety-guarded version of the same underlying model as the restricted Claude Mythos 5. It launched June 9, 2026, but was suspended on June 12 under U.S. export controls after Amazon researchers demonstrated a safeguard bypass, then redeployed globally on July 1 once controls were lifted. Anthropic has extended included access on all paid plans from July 7 to July 12, at up to 50 percent of weekly usage limits, after which Fable 5 moves to usage-credit pricing of $10 per million input tokens and $50 per million output tokens.

Key Points

The most capable AI model Anthropic has ever released to the public is free inside every paid Claude plan until July 12. That is the news. The reason it is news, and not just a line in a changelog, is that this is the third time in a month the answer to a simple question, can you actually use Fable 5, has changed.

Anthropic just pushed the deadline back five days, from July 7 to July 12. For a model that has spent most of its short life either unavailable or on a countdown, five more days is the whole story in miniature.

What Fable 5 Actually Is

Anthropic launched Claude Fable 5 on June 9, alongside a second model called Claude Mythos 5. They are the same underlying model. Mythos 5 is the raw version, handed only to authorized partners in a program called Project Glasswing and a handful of biology researchers. Fable 5 is that same model with safety guardrails bolted on and opened to everyone. The company points out that the name fits: Fable comes from the Latin fabula, "that which is told," which is close to the Greek mythos.

On capability, Anthropic did not hedge. Fable 5 is state of the art on nearly all the benchmarks it tested, and the harder and longer the task, the wider its lead. Stripe said the model "compressed months of engineering into days" on a migration across a 50-million-line codebase. It rebuilds working app code from a screenshot, holds focus across millions of tokens, and posts the company's highest scores on finance and science work.

The guardrails are the interesting part. Fable 5 runs classifiers for cybersecurity, biology, chemistry, and attempts to copy the model. When one trips, your request does not get refused, it quietly gets answered by the older Claude Opus 4.8 instead. Anthropic says that swap happens in under 5 percent of sessions.

Three Days, Then The Government Stepped In

Fable 5 was supposed to reach paid subscribers from June 9 to June 22. It lasted about three days.

On June 12, the U.S. government applied export controls and access was suspended for everyone. The trigger, by Anthropic's own account, was a piece of research from Amazon. Researchers found a way around Fable 5's safeguards, prompting it to hunt for software vulnerabilities, and in one case getting it to produce "code demonstrating how the relevant vulnerability could be exploited."

Anthropic pushed back on the severity. It said less capable models could produce the same output, and that the technique did not reveal anything unique to the Mythos-class model underneath. None of that mattered in the moment. The controls stayed on, and the best model on the market went dark for most of June.

The July 1 Comeback

The export controls were lifted on June 30. A day later, on July 1, Anthropic redeployed Fable 5 worldwide, back across Claude.ai, Claude Code, the Claude Platform, and Cowork, with the cloud providers switching it on again behind them.

The company also did something telling on the way back in: it opened a bug bounty on HackerOne specifically for people who can jailbreak Fable 5. Having just been pulled offline over exactly that kind of bypass, Anthropic is now paying the public to find the next one before a regulator does.

Now It Is A Countdown

Here is where the July 12 date comes in. Fable 5 is included in Pro, Max, Team, and select Enterprise plans, but only for up to half of your weekly usage limit, and only for now. The original plan cut that included access off on July 7. Anthropic has now extended it to July 12.

After that, Fable does not vanish, it changes price. Included access ends and the model moves to usage credits, meaning you pay per token to keep using it, at $10 per million tokens in and $50 per million out. Anthropic has said Fable is not leaving subscriptions permanently and will return as capacity allows, which is a polite way of saying availability depends on how much compute it has to spare.

Access Is The New Product

Strip out the model names and the Fable 5 saga is about something bigger than one release. The frontier of AI is now entangled with export policy, national security, and raw compute supply, and none of those move on a product schedule.

For a normal piece of software, "it is out" is a permanent state. For the most capable models, it is a window. Fable 5 was live for three days, pulled by the government, brought back, and is already on a clock that keeps getting nudged. What you are being offered until July 12 is not a product you own, it is a room you are allowed to stand in for a while.

If you want to know what the best AI can actually do right now, that is the argument for spending the next few days finding out. The window is open. It has a habit of closing.

Frequently Asked Questions

Until when is Claude Fable 5 free on paid plans?

Anthropic extended included access from July 7 to July 12, 2026. Fable 5 is included on Pro, Max, Team, and select Enterprise plans for up to 50 percent of your weekly usage limit through that date.

What happens to Fable 5 after July 12?

Included subscription access ends and Fable 5 moves to usage credits, meaning you pay per token, at $10 per million input tokens and $50 per million output tokens. Anthropic says Fable is not leaving permanently and will return as capacity allows.

What is the difference between Claude Fable 5 and Claude Mythos 5?

They are the same underlying model. Mythos 5 is the raw version restricted to authorized Project Glasswing partners and select biology researchers. Fable 5 is that model with safety guardrails added and released to the general public.

Why was Claude Fable 5 taken offline in June?

On June 12, the U.S. government applied export controls and suspended access after Amazon researchers demonstrated a way around Fable 5's safeguards, in one case getting it to produce code exploiting a software vulnerability. Anthropic said less capable models could produce the same output. Controls were lifted June 30 and Fable 5 returned July 1.

How capable is Claude Fable 5?

Anthropic says Fable 5 is state of the art on nearly all benchmarks it tested, with its lead growing on longer, harder tasks. Stripe said it compressed months of engineering into days on a 50-million-line codebase migration.

What are Fable 5's safety guardrails?

Fable 5 runs classifiers for cybersecurity, biology, chemistry, and model-copying attempts. When one triggers, the request is answered by the older Claude Opus 4.8 instead of Fable 5. Anthropic says this happens in under 5 percent of sessions.

Topics: ai-safety, llm, hackerone, anthropic, mythos-5, ai-models, claude-opus-4-8, stripe, claude, export-controls, claude-fable-5, usage-credits, frontier-ai, fable-5

More in tech