Claude Fable 5 is available again. Anthropic and the United States government announced that the advanced AI model returned to users worldwide starting Wednesday, July 1, following a short suspension tied to export control review. In addition to that, the model comes with a security classifier that should address the reported vulnerability, as well as some temporary usage restrictions which would later transition to a completely new price strategy.
Why Fable 5 Was Paused In The First Place
Before the current restoration of the functionality, the Fable 5 AI model was temporarily paused because of the export control issues. During this period, Anthropic collaborated directly with the US federal government in order to review the case and to understand what needs to be done in order for Fable 5 to become available again. That review apparently concluded quickly enough for a full relaunch this week, with the model now available to users around the world.
As for Mythos 5, which is another advanced AI model developed by Anthropic, it is getting reintroduced in a slightly different manner. Rather than restoring its worldwide availability right away, Anthropic decided to make the model available only to a certain list of organizations in the US for the moment.
A Public Nod From The Commerce Secretary
The United States Secretary of Commerce Howard Lutnick provided an official confirmation of the collaboration between Anthropic and the government in this matter via his social media page X. He said his office had spent the past two weeks working closely with Anthropic to review and approve Fable 5, describing the effort as a way to keep the US government aligned and strengthen the country’s position in AI.
As can be expected, however, this confirmation was not met with pure joy by everyone who uses Anthropic’s products. Many comments under the message of the Secretary expressed dissatisfaction with the amount of time the model was unavailable.
The Vulnerability Anthropic Says It Fixed
At the same time with the announcement about the relaunch of the model, Anthropic announced that it had addressed the mentioned above security vulnerability. According to the company’s announcement, it managed to develop a new classifier that was designed specifically in order to catch and block such actions, and it aims to catch the vast majority of attempts, which means that its effectiveness is expected to be at least 99 percent.
For those people who may not know about the term yet, it is important to explain that the classifier here is basically an automated screening tool. It analyzes a user’s prompt in order to determine whether the user made an ordinary request or tried to bypass the model’s safety measures in order to use its capabilities for something illegal. When a request gets flagged, the system doesn’t just block it silently. Anthropic said users will be notified, and the blocked request is automatically rerouted to Claude Opus 4.8 instead, so the person isn’t left without a response.
Anthropic Admits The Fix Isn’t Perfect Yet
As can be expected, Anthropic admits that there are some limitations of the classifier it has developed. Specifically, it warns that some benign requests can get blocked by mistake when the user performs some routine coding or debugging that involves technical jargon.
As Anthropic says, it will continue to improve the classifier in order to reduce the number of false positives. That kind of admission is worth noting, since it signals the company is treating this less as a finished solution and more as a first version that will need adjustment as real usage reveals where the system is overly cautious.
Being transparent about a security tool’s current limitations, rather than presenting it as flawless, is also fairly rare in the AI industry, where companies often stay quiet about the tradeoffs built into their safety systems.
New Usage Limits Take Effect Right Away
Access to Fable 5 isn’t unlimited at the moment. Usage is capped starting with the relaunch, and that cap will remain in place through July 7. After this day, the usage policy will change to a credit system.
These restrictions apply to several products by Anthropic, including Claude Platform, Claude.ai, Claude Code, and Claude Cowork. This temporary limit is available to people who use Pro, Max, Team, and some Enterprise plans.
Cloud platform access will be restored soon. Anthropic says that it will try to make Fable 5 available in AWS, Google Cloud, and Microsoft Foundry as quickly as possible, but no exact dates are provided at the moment.
Mythos 5 Access Is Expanding More Slowly
While the global availability of Fable 5 has been fully restored, Mythos 5 is slowly returning. Anthropic confirms that access to Mythos 5 was reinstated for a certain set of organizations in the United States following the government’s approval of June 26.
Moreover, Anthropic continues to collaborate with the government in order to expand the range of organizations and countries it can reach out to. Eventually, it expects to reach a wider range of partners from the so-called Glasswing program, but no concrete timeline was provided.
What This Means Going Forward
For everyday users, the biggest practical changes right now are the usage cap through July 7 and the possibility of hitting an occasional false flag from the new classifier, particularly during coding tasks. Anyone relying on Fable 5 for development work may want to keep that in mind over the next several days.
The Mythos 5 rollout looks more incremental, tied closely to ongoing coordination between Anthropic and federal regulators. Considering how rapidly the restoration of access to Fable 5 was organized, one can expect information regarding broader access to Mythos 5 within the coming weeks.
However, as of now, nothing has been officially planned or confirmed. The willingness of Anthropic to openly discuss both the fix for the security problem and its flaws, combined with the openness of the government regarding its participation in the process of evaluation, demonstrates that this is the situation when both sides conduct their affairs quite transparently despite dissatisfaction of certain users with the duration of the pause.
Frequently Asked Questions
Q1: Why was Claude Fable 5 unavailable before this relaunch?
Fable 5 was affected by a US export control review, and access was paused until Anthropic could solve the problem with the help of the federal government.
Q2: When did Fable 5 become available again?
Fable 5 became accessible to users worldwide beginning Wednesday, July 1, with the permission of the US government.
Q3: What security issue did Anthropic fix with this update?
Anthropic managed to address the vulnerability reported by Amazon by creating a new classifier to identify and block attempts at circumventing the model’s guardrails.
Q4: What happens if my request gets blocked by the new classifier?
If your request is blocked by the classifier, you will receive notification of it, and the request will automatically be rerouted to Claude Opus 4.8.
Q5: Does the new classifier ever make mistakes?
Yes. Anthropics admitted the possibility of flagging of some benign requests, especially when doing coding and debugging. The company also stated that they will continue to improve the classifier.
Q6: Are there usage limits on Fable 5 right now?
Yes. Until July 7, Fable 5 is available with up to 50 percent of the weekly usage limit, after which access will be limited to usage credits.
Q7: Which platforms currently support Fable 5?
Fable 5 is available through the Claude Platform, Claude.ai, Claude Code, and Claude Cowork for users on Pro, Max, Team, and selected Enterprise plans.
Q8: Is Fable 5 available on cloud platforms like AWS or Google Cloud?
Not yet entirely. Anthropic stated that they plan to resume access on AWS, Google Cloud, and Microsoft Foundry as soon as possible, although the exact timeline was not specified.
Q9: What is the status of Mythos 5?
Access to Mythos 5 has been restored to a specific set of organizations in the United States after government approval on June 26.
Q10: Will Mythos 5 access expand further?
Anthropic states that they plan to coordinate with the government to expand access to Mythos 5 for a broader range of domestic and international partners.
