AI Catchup

Anthropic Suspended Fable 5 and Mythos 5 on June 12, 2026, and Restored Them on July 1: What Happened

By 9 min read

A US export control directive forced Anthropic to disable Claude Fable 5 and Mythos 5 for all customers on June 12, 2026. The controls were lifted June 30 and Fable 5 returned globally July 1 with a new safety classifier. As of September 2026 both models are active, and Fable 5.1 is the current Fable model.

Status (October 9, 2026): over. Fable 5 and Mythos 5 have been back since July 1, 2026 and are listed as Active. Fable 5.1 is the current Fable model, and Anthropic now points most workloads to Opus 5.5.

On June 12, 2026 Anthropic suspended access to Claude Fable 5 and Mythos 5 for every customer to comply with a US government export control directive. The outage lasted 19 days. Anthropic says the export controls were lifted on June 30 and that Fable 5 was redeployed to users globally on July 1, with a new safety classifier that targets the bypass technique behind the directive. Mythos 5 access had already been restored to a set of US organizations after a government approval on June 26. On September 1 Anthropic shipped Fable 5.1 and Mythos 5.1, and its docs now call Fable 5 a legacy model that is still available.

This page is the record of the suspension and what followed. Every statement about the directive, the report behind it, and the safeguards is Anthropic's own characterization from its statement and redeployment posts.

Key Takeaways

  • Both models were disabled for everyone on June 12, 2026. The directive restricted access by any foreign national, inside or outside the United States. Anthropic says that because the order took effect immediately and it had no reliable way to verify nationality in real time, it suspended access for all users.
  • Other Claude models were never affected. Anthropic's statement said access to all other Anthropic models would not be affected, and the redeployment post does not report any change to that.
  • The controls were lifted on June 30 and Fable 5 returned on July 1. Anthropic says Fable 5 became available to users globally on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork, with AWS, Google Cloud, and Microsoft Foundry re-enabled "as quickly as possible".
  • Fable 5 came back with a stricter classifier. Anthropic says the new classifier blocks the reported technique in over 99% of cases, routes blocked requests to Opus 4.8 with a notice to the user, and flags benign coding requests more often than before.
  • The trigger was an Amazon research report. Anthropic says Amazon researchers found a way to prompt Fable 5 into identifying software vulnerabilities and, in one case, producing exploit code. Anthropic says the same results were reproducible on less capable models.
  • As of September 2026 both models are active. Anthropic's deprecation table lists claude-fable-5 and claude-mythos-5 as Active, retirement not sooner than June 9, 2027, and Fable 5.1 is the current Fable model.

Timeline

Date (2026)What Anthropic says happenedSource
June 9Fable 5 and Mythos 5 released. Mythos 5 went only to a small number of Project Glasswing partners.Redeployment post
June 12, 5:21pm ETExport control directive received. Anthropic disabled both models for all customers the same day.Statement
June 26US government approval to restore Mythos 5 to a set of US organizations.Redeployment post
June 30Export controls on Fable 5 and Mythos 5 lifted. Redeployment post published.Redeployment post
July 1Fable 5 available again to users globally on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork.Redeployment post (update)
July 2Anthropic publishes the classifier categories and a draft jailbreak severity framework, and opens a HackerOne program for Fable 5 cyber jailbreaks.Safeguards post
July 7End of the window in which Fable 5 was included in up to 50% of weekly usage limits on Pro, Max, Team, and select Enterprise plans; usage credits after that.Redeployment post
September 1Fable 5.1 and Mythos 5.1 released. Fable 5 becomes a legacy model, still available.Model docs

What Anthropic Said on June 12

Anthropic's statement said the US government, citing national security authorities, had issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. Anthropic said the net effect was that it had to abruptly disable both models for all customers to ensure compliance, and that access to all other Anthropic models would not be affected.

The statement said the government's letter gave no specific details of its national security concern, and that Anthropic's understanding was that the government believed it had become aware of a method of bypassing, or "jailbreaking", Fable 5. Anthropic said it had reviewed a demonstration of the technique being used to identify a small number of previously known, minor vulnerabilities, that other publicly available models could discover them without a bypass, and that it disagreed that a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. It closed by saying it believed the situation was a misunderstanding and was working to restore access as soon as possible.

What Anthropic Said on June 30

The redeployment post fills in the cause. Anthropic says the directive came after the government became aware of a report in which Amazon researchers had found a method of bypassing Fable 5's safeguards: prompting it so that it identified a number of software vulnerabilities, and in one case produced code demonstrating how one of them could be exploited. Anthropic says its own testing confirmed that many less capable models, including Claude Opus 4.8, GPT-5.5, and Kimi K2.7, could identify the same vulnerabilities, and that every model it tested could produce the same exploit demonstration, a list it gives as Claude Haiku 4.5, Sonnet 4.6, Opus 4.6, Opus 4.7, Opus 4.8, GPT-5.4, GPT-5.5, and Kimi K2.7. Anthropic's position is that the technique exposed no unique Mythos-level cyber capability and that the behavior was a borderline case for Fable 5's safeguards involving routine defensive work.

Anthropic says it nonetheless trained an improved safety classifier, working with the government, that targets and blocks the reported behavior. Per Anthropic, the specific technique is blocked in over 99% of cases; a blocked request notifies the user and is sent to Opus 4.8 instead; and researchers from the Department of Commerce's Center for AI Standards and Innovation (CAISI) tested both the prior and the new safeguards. Anthropic also says the new classifier flags benign requests more often during routine coding and debugging, and that it will keep refining it to reduce false positives.

The post also announces a draft jailbreak severity framework developed with Amazon, Microsoft, Google, and other Glasswing partners, and a set of commitments to the US government: expanded pre-release access for designated government partners on models that materially advance the frontier in national-security-relevant areas, rapid notification and sharing of new safeguards when significant jailbreaks are found, dedicated Anthropic teams and compute for joint research, and work toward a shared voluntary security and evaluation standard across frontier providers. The July 2 follow-up post describes the four categories Fable 5's classifiers sort cybersecurity requests into (prohibited use, high-risk dual use, low-risk dual use, and benign use) and opens a HackerOne program for reporting cyber jailbreaks in Fable 5.

Plan Access After the Restoration

Anthropic says that for Pro, Max, Team, and select Enterprise plans, Fable 5 was included for up to 50% of weekly usage limits through July 7, 2026, after which it is available via usage credits. Its footnote on Enterprise: standard Enterprise seats have no included Fable 5 allowance and get access only through usage credits; premium Enterprise seats had Fable 5 included through July 7, drawing from each member's seat usage, and after that need usage credits enabled. If credits are not enabled, users on those seats do not have access to Fable 5.

The June 23 subscription-credit change described in the launch coverage was overtaken by the suspension; the July 7 date above is the one that applied after the restoration.

Where Fable 5 Stands in September 2026

Anthropic's model overview lists Fable 5 under "Legacy models (still available)" and says that although it is still available, you should consider migrating to Fable 5.1 for improved performance. The model ID is unchanged, claude-fable-5, on the Claude API, Amazon Bedrock (anthropic.claude-fable-5), Google Cloud, Microsoft Foundry, and Claude Platform on AWS, so the cloud platforms that were "re-enabled as quickly as possible" in July now list it. Pricing is $10 per million input tokens and $50 per million output tokens, the same as Fable 5.1. The deprecation table lists claude-fable-5 and claude-mythos-5 as Active with retirement not sooner than June 9, 2027; claude-mythos-preview was deprecated on June 9, 2026.

For what changed in the September release, including the 75% cheaper cache reads and the narrower cyber safeguards, see the Fable 5.1 and Mythos 5.1 coverage. For the biology-safeguard changes Anthropic made to Fable 5 in August, see Fable 5's biology safeguards.

What Builders Should Do Now

  • Nothing about availability. Fable 5 and Mythos 5 are active. If you failed over to Opus 4.8 in June and never moved back, the decision now is Fable 5.1 versus Opus 5.5, not whether Fable 5 exists.
  • Expect classifier fallbacks on security-adjacent coding. Anthropic says the post-restoration classifier flags benign coding and debugging requests more often and routes blocked Fable 5 requests to Opus 4.8 with a notice. Build for that fallback rather than around it. The Opus 4.8 Fast mode coverage describes the model your blocked requests land on.
  • Check your Enterprise seat type. Standard Enterprise seats need usage credits enabled to reach Fable 5 at all, per Anthropic's footnote.
  • Move off Fable 5 for new work. Anthropic's docs recommend migrating Fable 5 to Fable 5.1 at the same price; the migration notes cover the breaking changes. For most workloads, Anthropic's models overview now says to start with Opus 5.5 and reserve Fable 5.1 for demanding reasoning and long-horizon agentic work.

FAQ

For quick answers on current availability, why the models were suspended, what changed on the return, and whether to move to Fable 5.1, see the structured FAQ in this article's metadata. It mirrors the sections above in short form.

Corrections and Updates

  • October 9, 2026: This page said builders who stayed on Opus 4.8 were choosing between Fable 5.1 and Opus 5; Opus 5.5, released September 22, 2026, is now the current Opus model, and Anthropic's models overview recommends it for most workloads.

Sources

Read next

Related coverage

Frequently Asked Questions

Are Fable 5 and Mythos 5 available now?

Yes. Anthropic says the export controls were lifted on June 30, 2026 and Fable 5 returned to users globally on July 1 on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork. As of September 2026 Anthropic's deprecation table lists claude-fable-5 and claude-mythos-5 as Active, with retirement not sooner than June 9, 2027.

Why were Fable 5 and Mythos 5 suspended in June 2026?

Anthropic says the US government issued an export control directive on June 12, 2026 restricting access by any foreign national, inside or outside the United States. Because the order took effect immediately and Anthropic had no reliable way to verify nationality in real time, it suspended both models for all users. Other Claude models were not affected.

What changed when Fable 5 came back?

Anthropic says it trained an improved safety classifier, working with the government, that blocks the technique in the Amazon report in over 99% of cases. A blocked Fable 5 request notifies the user and is sent to Opus 4.8 instead. Anthropic also says the classifier flags benign coding requests more often than before.

Should I still use Fable 5, or move to Fable 5.1?

Anthropic's model docs list Fable 5 as a legacy model that is still available and recommend migrating to Fable 5.1, released September 1, 2026, at the same $10 per million input and $50 per million output price. Fable 5 keeps its model ID, claude-fable-5, on the Claude API, Bedrock, Google Cloud, and Foundry.

What reason did the government give for the directive?

Anthropic says the directive cited national security authorities without specific details, and that it followed a report in which Amazon researchers found a way to prompt Fable 5 into identifying software vulnerabilities and, in one case, producing exploit code for one of them. Anthropic says less capable models, including Opus 4.8 and GPT-5.5, could do the same.

Get the weekly AI Catchup

Tools, practices, and what matters, in your inbox every week.