Claude Fable 5 is back. The U.S. government has lifted export controls on Anthropic’s most powerful public AI model. Starting today, July 1, 2026, Fable 5 is once again available worldwide via the Claude platform, Claude.ai, Claude Code, and Claude Cowork. The more restrictive version, Claude Mythos 5, has also been made available to some users. This brings to an end a nearly three-week dispute between Anthropic and the White House, which had taken both models completely offline in June.
Key Points at a Glance
- On June 30, 2026, the U.S. Department of Commerce lifted export controls on Claude Fable 5 and Mythos 5.
- Fable 5 will return worldwide on July 1, initially with a limited usage quota until July 7.
- The suspension was triggered by a method reported by Amazon that circumvented Fable 5’s safeguards.
- Anthropic has introduced a new classifier that, according to the company, blocks the disputed technique in over 99 percent of cases.
- Pricing remains unchanged at $10 per million input tokens and $50 per million output tokens.
What happened?
On the evening of June 30, 2026, Anthropic announced that the U.S. Department of Commerce had rescinded the export controls on Claude Fable 5 and Claude Mythos 5. Commerce Secretary Howard Lutnick communicated the decision in a letter. Shortly thereafter, the company announced that it would gradually restore access starting the following day. Prior to this, both models had been blocked for all customers for about three weeks.
The conflict had been building for several weeks. This overview shows the key milestones:
| Date | Event |
|---|---|
| June 9, 2026 | Anthropic releases Claude Fable 5 and Mythos 5 as its most powerful models to date. |
| June 12, 2026 | The Department of Commerce issues an export control directive. Anthropic takes both models offline worldwide. |
| June 26, 2026 | Mythos 5 is made available again to select U.S. organizations. |
| June 30, 2026 | Export controls are completely lifted. |
| July 1, 2026 | Fable 5 returns worldwide. |
Why was Claude Fable 5 blocked in the first place?
The directive issued on June 12 prohibited foreign nationals from accessing Fable 5 and Mythos 5, explicitly including foreign Anthropic employees within the United States. Since the nationality of individual users cannot be verified in real time, Anthropic completely disabled both models for all customers rather than risk violating the directive.
According to Anthropic, the issue stemmed from a report by Amazon researchers. They had discovered a method to circumvent Fable 5’s protective mechanisms: The model could be instructed to identify software vulnerabilities; in one instance, it even generated code that demonstrated how to exploit a vulnerability. The government cited national security concerns.
Anthropic took a much more cautious view of the incident. According to its own tests, even weaker models such as Claude Opus 4.8, GPT-5.5, and Kimi K2.7 were able to detect the same vulnerabilities. All tested models reproduced the exploit proof, including the smaller Claude Haiku 4.5 and Sonnet 4.6. The technology in question thus did not reveal any unique capabilities of the Mythos class.
These new security mechanisms are part of the comeback
The return of Claude Fable 5 is subject to stricter conditions. In a letter to Anthropic, Lutnick noted that the company had committed to proactively identifying and addressing security-related risks, working with the government on release protocols for future models, and reporting detected malicious activities.
Technically, Anthropic is relying on a revised safety classifier. It is designed to block the technique described in the Amazon report in over 99 percent of cases. On the downside, it is also likely to falsely flag more harmless coding and debugging requests in the future. Fable 5 continues to forward problematic inputs to the less powerful Opus 4.8. In addition, Anthropic is setting up a team to continuously monitor reporting channels for jailbreaks and is launching a new HackerOne program through which security researchers can submit potential vulnerabilities. Together with Amazon, Microsoft, and Google, the company is also working on a standardized framework for assessing the severity of such bypasses.
We’ve already described in detail just how powerful the Mythos class is and how Fable 5 and Mythos 5 differ from one another when both models were first launched.
Availability and Pricing of Claude Fable 5
Starting July 1, Fable 5 will once again be available to all users worldwide on the Claude platform, Claude.ai, Claude Code, and Claude Cowork. Access via AWS, Google Cloud, and Microsoft Foundry is expected to follow as soon as possible. Terms and conditions vary depending on the respective plan:
| Plan | Access to Fable 5 |
|---|---|
| Pro, Max, Team, select Enterprise plans | Included until July 7 for up to 50% of weekly usage limits; thereafter via usage credits. |
| Standard Enterprise Seats | No included allocation. Usage is billed immediately via Usage Credits. |
| Premium Enterprise Seats | Included in the subscription through July 7; after that, usage is billed via Usage Credits. |
Pricing remains unchanged: via the API, Fable 5 continues to cost $10 per million input tokens and $50 per million output tokens. This makes it roughly twice as expensive as Opus 4.8. Fable 5 is also available to some U.S. organizations whose access was approved by the government on June 26. Anthropic plans to further expand the circle of eligible partners both domestically and internationally through its cybersecurity program, Glasswing.
Context: What This Case Reveals About AI Regulation
This episode illustrates just how closely Washington is now monitoring advanced AI models. The government considers Anthropic a supply chain risk and subjects it to export controls typically reserved for military equipment. Its competitor, OpenAI, complied with a similar request and is currently releasing its new flagship model, GPT-5.6, only to a select group of partners.
Criticism focused primarily on the timing: While the most powerful U.S. models were offline, Chinese open-source models gained time to catch up. On the night before the suspension was lifted, Anthropic had also unveiled its new mid-range model, Claude Sonnet 5, which comes close to matching the performance of Opus 4.8. It remains to be seen whether Fable 5 will remain usable in the long term without additional identity verification. Hints of possible KYC verification had recently appeared on Anthropic’s website, but there is no reliable confirmation yet that it will be mandatory for Fable 5.
Frequently Asked Questions About Claude Fable 5
Is Claude Fable 5 available again?
Yes. Since July 1, 2026, Fable 5 has been accessible again worldwide via the Claude platform, Claude.ai, Claude Code, and Claude Cowork, following the U.S.’s lifting of export controls.
Why was Fable 5 blocked?
An export control directive dated June 12, 2026, prohibited access by foreign nationals. Since Anthropic cannot verify users’ nationality in real time, the company disabled Fable 5 and Mythos 5 for all customers.
How much does Claude Fable 5 cost?
Via the API, Fable 5 continues to cost $10 per million input tokens and $50 per million output tokens. In Pro, Max, Team, and select Enterprise plans, it is included for up to 50 percent of weekly usage limits through July 7, 2026.
What is the difference between Fable 5 and Mythos 5?
Both share the same base model. Fable 5 is the publicly available version with additional safeguards. Mythos 5 is more restrictive and is reserved for a select group of organizations.
Conclusion
The return of Claude Fable 5 brings an end to an unusual regulatory dispute that took Anthropic’s most powerful public model off the market for nearly three weeks. For users, the most important thing is that the model is available again, albeit initially with a capped quota and a stricter safety classifier, which may occasionally be overly cautious when handling coding tasks. More interesting than the technical side is the political one: The case makes it clear that the availability of cutting-edge AI today also depends on regulatory decisions that can take effect within hours.
Further details are provided in Anthropic’s official announcement.
