Starting July 8, Claude Fable 5 is no longer included in Claude Pro, Max, Team, or select Enterprise subscriptions — it now requires usage-credit billing. If you want to keep using Fable 5, enable usage credits in your claude.ai account under Settings → Usage and set a spending cap to avoid surprise charges. Sonnet 5 and other models remain in your subscription at no extra cost. The Anthropic blog post confirmed the shift is not permanent.
What Changes on July 8
Through July 7, Fable 5 was included at up to 50% of weekly usage limits for subscribers — a temporary window Anthropic created when it restored the model on July 1 after US export controls were lifted. Starting July 8, that window closes. Fable 5 shifts entirely to usage-credit billing at Anthropic’s standard API rate: $10 per million input tokens and $50 per million output tokens.
To continue using Fable 5, you need to take one action: log in to claude.ai, go to Settings → Usage, and enable usage credits. Once enabled, you can also set a monthly spending cap — a useful safeguard if you use Fable 5 heavily for tasks like long-document analysis or coding projects. Without enabling usage credits, Fable 5 simply won’t appear as an option in the model selector.
| Access Type | Fable 5 Available? | Cost |
|---|---|---|
| Pro / Max / Team subscription | No (from July 8) | Usage credits: $10/1M input, $50/1M output |
| Enterprise (consumption-based) | Yes | API rate |
| Claude API direct | Yes | Standard API pricing |
Why Anthropic Is Doing This
The short answer is demand. Anthropic said in its original redeployment post that Fable 5 usage would be “very high, and difficult to predict,” so it opted for a staged rollout rather than immediately flooding the model into unlimited subscription access. The company has been aggressively expanding infrastructure — including a deal with SpaceX to access over 220,000 Nvidia GPUs at the Colossus 1 data center — but Fable 5’s capacity requirements are outpacing even that build-out pace.
The pattern looks a lot like what GitHub Copilot did with premium model access earlier this year: subscription tiers handle baseline models, while the most capable models shift to metered billing until the infrastructure catches up. For subscribers, this creates a two-tier reality: casual use of Sonnet 5 stays free, while power users who need Fable 5 for serious workloads pay proportionally more.
Anthropic Says Fable 5 Will Come Back to Subscriptions
The concern among Claude subscribers is obvious — and Anthropic moved quickly to address it. A Claude Code lead engineer, Thariq (@trq212), clarified on X on July 2 that the move off subscriptions is explicitly not a permanent decision. “We aim to restore Fable as a standard part of our subscriptions as soon as capacity allows,” he wrote. Anthropic’s own blog post said the same.
That’s a meaningful distinction. The earlier paywall that went into effect in June was also framed as temporary, and the July 1 restoration proved Anthropic follows through on that framing. The current shift is a capacity management measure, not a permanent tiering decision. There is no stated timeline for when Fable 5 will return to standard plans.
Frequently Asked Questions
Does Claude Fable 5 still work on the API after July 8?
Yes. The July 8 change only affects subscription-based access (Pro, Max, Team, Enterprise). Direct API users continue accessing Fable 5 at standard API pricing with no interruption.
How much will Fable 5 cost me per month with usage credits?
It depends entirely on your usage. At $10/1M input tokens and $50/1M output tokens, a typical long-form conversation (roughly 10,000 input + 2,000 output tokens) costs about $0.20. Heavy power users running dozens of complex sessions daily could see $20–50/month in credit charges on top of their subscription.
Will my existing Claude subscription price change?
No. Subscription pricing for Pro, Max, and Team plans is unchanged. Usage credits are an optional add-on layer — you only pay extra if you opt in and actually use Fable 5.
Is Sonnet 5 as good as Fable 5?
For most everyday tasks — writing, summarizing, coding assistance — Sonnet 5 performs very well and remains included in all subscription tiers. Fable 5 outperforms it on complex reasoning, long-context tasks, and coding benchmarks, which is why it commands higher infrastructure costs.
