Why is Claude Fable 5 unavailable? The June export-control ban, its return, and what you can use now

"Why is Claude Fable 5 unavailable?" is still one of the top suggestions when you type the model's name into Google, months after the event that started it. On June 12, 2026, three days after Anthropic released Fable 5, the US Commerce Department ordered the company to keep foreign nationals from using it. Anthropic switched the model off for every Claude user worldwide. The ban was lifted on June 30, and Fable 5 came back on July 1. If you can't use it today, the export controls aren't the reason. Your plan almost certainly is.

This explainer is based on Anthropic's statements and documentation, the Claude Help Center, and reporting by Bloomberg, Reuters (as cited by other outlets), Export Compliance Daily and The Next Web, checked October 16, 2026.

The short version

• What happened: Fable 5 launched on June 9, 2026. On June 12, a Commerce Department export-control directive required Anthropic to cut off access for any foreign national, inside or outside the US. Anthropic had no reliable way to check nationality in real time, so it disabled Fable 5 and Mythos 5 for everyone.
• Why: Anthropic understood the order to relate to a jailbreak reported by Amazon researchers. They had found a way to get Fable 5 to find software vulnerabilities, once with exploit demo code.
• The return: Anthropic says the controls were lifted on June 30. Fable 5 was back worldwide on July 1, with a new safety classifier.
• Today: Fable 5 and its successor, Fable 5.1 (September 1, 2026), are on all paid Claude plans and the API. On Pro and standard Team seats they run on pay-as-you-go usage credits, which is the most common reason the model looks "unavailable."

What happened to Claude Fable 5?

Anthropic released Claude Fable 5 and Claude Mythos 5 on June 9, 2026. They are the same underlying model. Fable 5 shipped with strong safeguards for general use. Mythos 5, with fewer safeguards, went only to Project Glasswing partners doing defensive cybersecurity work.

On Friday, June 12, Anthropic said it had received an export-control directive at 5:21 p.m. ET. The directive suspended all access to both models "by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees." To comply, Anthropic disabled both models for all customers. It said other models, including Claude Opus 4.8, were unaffected.

Anthropic said the government had so far given it only "verbal evidence of a potential narrow, non-universal jailbreak," and that it disagreed "that the finding of a narrow potential jailbreak should be cause for recalling a commercial model." It complied anyway.

Other reporting fills in the mechanism:

• A US official confirmed on June 13 that the Commerce Department had issued the directive, according to Reuters as cited by the National Law Review.
• Export Compliance Daily described the June 12 letter as an "is-informed" letter, a notice that tells a company a license is now required for certain exports. Bloomberg described it as an export-control rule imposed through a private letter.
• Greenberg Traurig's lawyers, writing in the National Law Review on June 17, called it the first known US use of export-control authority to regulate a specific frontier AI model on national security grounds.

The Next Web reported on June 24 that Legion LegalTech, a San Jose legal-software company with developers in Canada, had sued in federal court in Washington to have the directive vacated.

Why was Fable 5 banned?

Anthropic's June 30 post, "Redeploying Claude Fable 5," gives its account. The government had become aware of a report from Amazon researchers who found a way to bypass Fable 5's safeguards by asking it to identify software vulnerabilities. In one case, the model produced code demonstrating how a vulnerability could be exploited.

Anthropic says its own testing found that less capable models could find the same vulnerabilities and write the same demonstration, so the technique exposed no unique Mythos-level cyber capability. That is Anthropic's view; the government hasn't published its assessment.

How did Fable 5 come back?

According to Anthropic's June 30 post, updated July 1:

1. A new classifier. Anthropic trained an improved safety classifier, built with the government, that it says blocks the reported technique in over 99% of cases. The Commerce Department's Center for AI Standards and Innovation (CAISI) tested the old and new safeguards. When a request is blocked, you're told, and it's sent to Opus 4.8 instead. Anthropic acknowledges the classifier also flags harmless coding and debugging requests more often.
2. Mythos first. On June 26, the government approved restoring Mythos 5 for a set of US organizations. Export Compliance Daily reported that Commerce Secretary Howard Lutnick's letter said Commerce could adjust the scope "should circumstances change."
3. Controls lifted. Anthropic says the export controls on both models were lifted as of June 30. Bloomberg reported on July 1 that the government had lifted restrictions on foreign access to Fable 5.
4. Back online. Fable 5 became available worldwide on July 1 on the Claude Platform, claude.ai, Claude Code and Cowork. Anthropic's API release notes for that day record that access to both models was restored. Amazon said the same day that Fable 5 was back on Amazon Bedrock.

Why does Claude say Fable 5 is unavailable now?

As of October 16, 2026, Anthropic's model deprecations page lists claude-fable-5 as Active, with retirement "not sooner than June 9, 2027." If you can't use it, check these causes, in order of how often they come up:

• You're on Pro or a standard Team seat without usage credits. The Help Center says Fable 5 and 5.1 aren't included in these plans' limits and run on pay-as-you-go usage credits from the start. This changed on July 20, 2026, when a launch promotion ended (Anthropic's June 30 post said it would run through July 7; the Help Center gives July 19 as the end). Pro accounts got a one-time $100 credit, claimable until August 2. Those credits expired on September 17, 2026.
• You're on a standard seat-based Enterprise plan. Fable isn't in plan limits, and your organization has to enable usage credits.
• Your organization hasn't enabled Fable models. The Help Center lists this among the reasons a Team or Enterprise user can't see Fable 5 or 5.1.
• You've hit the Fable cap on Max or a premium seat. These plans can spend up to 50% of their weekly limit on Fable models. After that, you need usage credits or another model.
• You're on the Free plan. Fable models are on paid plans only.
• Your Claude Code or Desktop app is too old. The Help Center lists Claude Code version 2.1.170 or later for Fable 5 and 2.1.255 or later for Fable 5.1, and the latest Claude Desktop for Fable in Cowork.
• Your organization uses zero data retention. Anthropic's release notes say Fable 5 requires 30-day retention and isn't available under ZDR. Its Fable 5.1 announcement adds that eligible customers can use Fable 5 with ZDR until its Enterprise Frontier Safeguards option is ready, so ask your account team.
• A request was blocked, not the model. A classifier refusal returns stop_reason: "refusal" in the API, or a notice in the apps, and may be re-run on another model.
• A temporary outage. Anthropic's status page has logged short incidents for Fable 5.1, most recently "Elevated errors for Claude Mythos 5.1 and Claude Fable 5.1" on October 5, since resolved. It showed all systems operational on October 16, 2026.

There's also a known bug. In a GitHub issue (checked October 16, 2026), Max users reported that Claude Code's /model command said Fable "runs on usage credits" and blocked the switch, even though /usage showed their included Fable allowance mostly unused. No Anthropic response appears in the thread. Several users said passing --model claude-fable-5-1 at launch worked. These are user reports, not confirmed by Anthropic.

Is identity verification required to use Fable 5?

No, and Anthropic says the two aren't connected. The June episode exposed that Anthropic had no real-time way to check a user's nationality, and in the same month it updated its privacy policy to say: "In certain circumstances, we may ask you to verify your age or identity." Techzine (June 16) and Tech Times (June 21) reported that the change took effect on July 8, 2026, and linked it to the Fable shutdown. Anthropic disputes that link. TechCrunch reported on June 22 that Anthropic pointed to a post by its Thariq Shihipar saying the policy was updated "as an update to the appeals process," for accounts flagged for possible fraud, and that "It's unrelated to the Fable or Mythos rollout."

Anthropic's Help Center article on identity verification, run through Persona, doesn't mention Fable or export controls. Our story on Claude's 18+ age checks covers both kinds of verification.

Claude Fable 5 vs Fable 5.1: what's the difference?

Anthropic released Claude Fable 5.1 on September 1, 2026, and calls it Fable 5's successor. It is generally available on the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud and Microsoft Foundry. Anthropic says it is better at coding and knowledge work and fires safety interventions less often: about 60% fewer cyber interventions per Claude Code session, and 85% fewer biology triggers on harmless questions than at Fable 5's launch.

Last verified: October 16, 2026 (Anthropic pricing and model deprecations pages)

Anthropic estimates the cheaper cache reads make typical workloads about 25% cheaper on Fable 5.1 than on Fable 5. Fable 5.1 also introduced preserved thinking, part of the anti-copying measures in our distillation defenses explainer.

Claude Fable vs Opus: which should you use?

Anthropic says Claude Opus 5.5 (September 22, 2026, $4/$20 per million tokens) "performs at the level of Claude Fable 5.1 on most work." Those are Anthropic's own comparisons. Its guidance is to move to Fable 5.1 only when Opus 5.5 at higher effort still falls short on your own evaluations. Our guide to which Claude model to use has the full lineup and prices.

Will Fable 5 come back, or go away?

Fable 5 is already back. The open question is how long it stays. Anthropic hasn't announced a deprecation, and it says it gives at least 60 days' notice before retiring a public model. With Fable 5.1 at the same price and cheaper cache reads, there's little reason to start new work on Fable 5.

The government's power to act again hasn't gone away. Lutnick's letter reserved the right to adjust the scope of license requirements, and a White House official told Semafor in October that export controls could be one tool for enforcing new AI safety rules, as our story on the White House Super Intelligence Force reports. Anthropic's company explainer covers the wider relationship with Washington.

What's confirmed, reported and uncertain

• Confirmed (Anthropic): the June 9 launch; the June 12 directive and the global shutdown; Anthropic's disagreement; the Amazon jailbreak report; the new classifier and Opus 4.8 fallback; Mythos 5 restored for US organizations on June 26; controls lifted June 30; global return July 1; Fable 5's active status and retirement floor; plan and credit rules; Fable 5.1's release, prices and availability.
• Reported: a US official's June 13 confirmation (Reuters, via the National Law Review); the "is-informed" letter and Lutnick's language (Export Compliance Daily); Bloomberg's July 1 report on the lifting; the Legion LegalTech lawsuit (The Next Web); the July 8 effective date for the privacy-policy change (Techzine, Tech Times); Anthropic's statement that identity checks are unrelated to Fable (TechCrunch).
• Uncertain: the government's own assessment of the jailbreak; whether the identity-verification policy will ever be used for nationality checks; the status of Legion's lawsuit; and when Fable 5 will be retired.