Dario and Daniela Amodei's Anthropic extended promotional access to Claude Fable 5 for paid Claude users through July 12th, the Claude account said in a two-post thread on X on Tuesday.
The extension gives subscribers five more days before Fable 5 stops being treated like a partially bundled subscription model. Anthropic said users can still spend up to 50% of their weekly usage limit on Fable 5. Once they hit that cap, they can either keep using Fable 5 with usage credits or switch to another Claude model and continue working within their remaining plan limits.
That detail is the commercial hinge in the announcement. Fable 5 is Anthropic's highest-end public model, and Anthropic is letting paid users sample it inside existing subscriptions only within a capped allowance. The company is preserving access for users who have built workflows around Fable since its July 1st restoration, while steering heavy users toward a metered billing system.
Anthropic's linked help-center article describes the original promotion as available on Pro, Max, Team and premium seats on seat-based Enterprise plans, where enabled by an organization. The same page says Free users are excluded, as are standard seats on legacy seat-based Enterprise plans, usage-based Enterprise plans and API usage. API access to Fable 5 is billed separately at standard rates.
The help page had described the promotional window as running from July 1st through July 7th at 11:59:59 PM PT. Tuesday's X post moves the practical cutoff for paid-plan promotional access to July 12th, while leaving the usage structure unchanged: Fable 5 draws from the regular weekly plan limit and uses that limit faster than other Claude models, according to Anthropic's support language.
The timing matters because Fable 5's first month has been defined as much by access rules as by model capability. Anthropic released Fable 5 and Mythos 5 on June 9th, describing Fable 5 as its most capable widely released model for demanding reasoning and long-horizon agentic work. Mythos 5 shares the same capabilities, according to Anthropic's docs, but is limited to approved customers in Project Glasswing.
Three days after launch, on June 12th, the U.S. government applied export controls to Fable 5 and Mythos 5. Anthropic said it had no reliable way to verify users' nationality in real time, so it suspended both models for all users. On June 30th, Anthropic said those controls had been lifted, and the company restored Fable 5 globally on July 1st across Claude Platform, Claude.ai, Claude Code and Claude Cowork.
The Associated Press reported that the Commerce Department's restrictions followed cybersecurity concerns and that Anthropic traced the episode to a report from Amazon researchers who had found a way to bypass Fable 5's safeguards. Anthropic's own redeployment post said the incident showed the industry needs a consistent framework for assessing and fixing jailbreaks.
That is why this subscription extension is more than a customer-service patch. The Amodeis founded Anthropic around the claim that frontier models should be deployed with stronger safety systems and tighter controls. Fable 5 is where that thesis meets the pricing page: the model is available broadly enough to become part of everyday paid Claude workflows, but constrained enough that Anthropic can meter expensive or risky use.
For developers, the constraints are explicit. Anthropic's docs say Fable 5 uses the API model ID claude-fable-5
, supports a 1 million token context window by default and up to 128,000 output tokens per request, and is priced at $10 per million input tokens and $50 per million output tokens. The docs also say Fable 5 includes safety classifiers that can decline certain requests, with fallback handling available for integrations.
For subscription users, the immediate rule is simpler: paid access continues through July 12th, but only up to half of the weekly usage limit. After that, the same model becomes a credits decision.