Claude Fable 5's Free Window Ends While Access Is Still Suspended

The Free Window Ends, But Access Is Still Paused

Anthropic launched Claude Fable 5 and Claude Mythos 5 on June 9, 2026, positioning Fable 5 as the most capable model it had made generally available. To let developers try it without a separate bill, Anthropic included Fable 5 on Pro, Max, Team, and seat-based Enterprise plans at no extra cost for a two-week window that ended on June 22.

There is a catch: Anthropic's separate access statement says Fable 5 and Mythos 5 were removed from all users after a US government directive, while Anthropic works to restore access. As of June 28, that means the June 23 change is best read as a billing and plan-entitlement change, not as a guarantee that the model is usable. If access is restored after the free window, continued use inside a subscription plan requires usage credits instead of being bundled in for free.

This deadline mainly affects developers who were evaluating Fable 5 through a claude.ai subscription before access was paused. API and cloud availability is also subject to the suspension; when available, Fable 5 is priced at the standard Fable rate rather than being covered by the subscription window.

Pricing: Double Opus 4.8

Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, with the standard prompt-caching discount on cached input tokens still available. Anthropic has said it intends to bring Fable 5 back into subscription plans once it has more inference capacity, but hasn't given a date for that - and the June 12 availability suspension adds a separate uncertainty.

What Developers Found Out During the Free Window

The brief free-access period still turned into an unplanned stress test of token budgets, though the evidence is anecdotal. Developers Digest cites one user who hit the $200-a-month Max plan's limits in under 30 minutes and heavy users reporting individual tasks of 500K to 1M tokens. It also cites a single pre-launch tester who found Fable 5 used roughly half the tokens of Opus 4.8 for the same agentic work. Anthropic's launch post says only that Fable 5 is more token-efficient than past Claude models, without a figure. If a ratio like that holds on your own workloads, it narrows - but doesn't close - the 2x price gap once usage credits are involved.

A Separate Complication: Export Controls

Three days after launch, on June 12, Anthropic removed access to Fable 5 and Mythos 5 following a US government export-control directive. The directive targeted access by foreign nationals, and Anthropic said complying meant disabling both models for all customers, disrupting availability before most teams had finished a first integration. That's a different story from the pricing change - one we cover in When Your AI Model Gets Pulled - but the two events compound: the same model was hit by both an availability shock and a cost shock inside its first two weeks of public life.

What This Means for Your Setup

• If you tested Fable 5 for free, check your config before access returns. Anything still pointed at Fable 5 may draw on usage credits if Anthropic restores the model after the free window - confirm that's intentional, not a surprise on next month's invoice.
• Run the comparison before you pay for it. At Fable's premium token price, it is only worth keeping for workloads where it measurably outperforms Opus 4.8 on your own evals, not because it benchmarks higher in general.
• Don't hardcode the model you tested for free. This is exactly the failure mode described in When Your AI Model Gets Pulled: a model's availability and price tier have both changed inside two weeks. Keep the model ID in one config value and a fallback validated against Opus 4.8.
• Track the cost, not just the capability. If Fable 5 earns its place in a production workflow, pair it with the practices in our AI cost modeling guide before usage credits turn into a surprise bill.