Claude Fable 5 Is Free Until 22 June and Here Are 7 Things You Should Try Before Anthropic Starts Charging

Anthropic has given paying Claude users its most powerful artificial intelligence (AI) model yet, and for the next 13 days, it costs nothing extra to use.

The company launched Claude Fable 5 on 9 June 2026, the first public version of its Mythos-class technology, a tier it had kept out of public hands over cybersecurity fears. The model is free on Pro, Max, Team, and seat-based Enterprise plans through 22 June. From 23 June, using it needs usage credits priced at $10 (£7.47) per million input tokens and $50 (£37.35) per million output tokens, double the cost of the firm's Opus 4.8 model.

Anthropic says Fable 5 outperforms any model it has made generally available. Here are seven things to try before the free window shuts.

Let It Code for Hours on Its Own

Fable 5 can run complex coding tasks for hours without losing track of the job. Payments firm Stripe found it completed a migration across a 50-million-line code base in one day, work that would have taken a team more than two months by hand.

Turn a Screenshot Into a Working App

The model can rebuild a web app's underlying code from screenshots alone, with no access to the original files. A rough picture of an app can become a working prototype, even for people who cannot code.

Run Senior-Level Financial Analysis

Anthropic calls Fable 5 its strongest finance-focused model so far. Analytics firm Hex reported it became the first model to pass 90% on its benchmark of complex, long-running analytical tasks, a 10-point jump over Opus 4.8, as covered in VentureBeat.

Speed Through Dense Research

An outside physics research group found that Fable 5 reached in 36 hours roughly what a rival model produced over four days, while using a third of the reasoning steps. For anyone sifting through heavy data or technical reports, that changes what a single afternoon can deliver.

Watch It Finish a Pokémon Game

In a lighter test, Fable 5 played through the Game Boy Advance title Pokémon FireRed from start to finish using only raw screenshots and a minimal set of tools. Earlier Claude models could not beat the game even with extra help.

Let It Check Its Own Work

Anthropic says Fable 5 writes its own tests, checks its output against the goal it was given, and refines the result before handing it back. Shopping rewards platform Rakuten said this self-checking is what makes highly autonomous work possible.

Pull Data Out of Messy PDFs

Its vision skills let the model read diagrams, charts, and tables buried inside PDF files. That matters for legal, finance, and architecture work, where the key detail often sits in the layout rather than the text.

Guardrails and Deadlines

The catch is the deadline. Anthropic expects demand to run high and is releasing access in stages, which is why the free period ends on 22 June before credits take over.

The model also ships with guardrails. Requests involving cybersecurity, biology, chemistry, or attempts to copy the model are blocked and routed to Opus 4.8 instead. Anthropic says fewer than 5% of sessions trigger that fallback, and that more than 1,000 hours of outside testing found no reliable way around its safety classifiers.

For subscribers, the takeaway is simple. The most capable AI Anthropic has put in public hands is free until 22 June, and the next 13 days are the cheapest chance to find out what it can do.