Anthropic’s Mythos 5 is moving back into use, but only in a narrow way. After two weeks of negotiations with the Trump administration, the company has received approval to redeploy the model to a select group of organizations, while Fable 5 remains unresolved.
What Changed For Mythos 5
The shift comes from a letter dated June 26th, sent by Commerce Secretary Howard Lutnick to Anthropic co-founder Tom Brown. According to the source, Tom Brown had recently been leading negotiations with the government.
The letter says there has been a “revision to the license requirements” after Anthropic “worked with the U.S. government to address risks” tied to Mythos 5 and Fable 5. That does not mean the broader restrictions have disappeared. It means Mythos 5 now has a limited path back for approved users.
Anthropic spokesperson Danielle Ghiglieri said the company had “received notice from the US government that Mythos 5, our strongest cybersecurity model, can be redeployed to a small group of cyber defenders and infrastructure providers.” She also said Anthropic is working to provision the approved providers and restore access as quickly as possible.
The practical result is a partial reopening. Mythos 5 is not broadly available again, but the government has cleared specific organizations to use it under an exception.
Fable 5 Is Still Waiting
The same approval does not appear to apply to Fable 5, described in the source as the public-facing Mythos-class model. Fable 5 remains in limbo, and the source says there is no apparent timeline for a rollout agreement.
That distinction matters because the two models are being treated differently. Mythos 5 is returning for a small group tied to cyber defense and infrastructure. Fable 5, by contrast, is still not available for general use again.
Danielle Ghiglieri said Anthropic continues to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again. For now, however, the update is limited to the approved Mythos 5 group.
The Export Control Directive Remains
The US government did not lift the export control directive that it placed on Anthropic two weeks ago. That directive barred any foreign national from accessing either model, including Anthropic’s own employees.
Instead of removing the directive, the government created an exception for Mythos 5. The source compares the structure to the approval given for OpenAI’s GPT-5.6, which was announced earlier today.
Under this exception, Anthropic employees who are not US nationals are cleared to access Mythos 5. Members of the approved organizations who are not US nationals are also cleared, according to the letter.
Lutnick wrote that the work so far had produced “significant progress.” He also wrote that Anthropic had committed to work with the U.S. government on protocols, standards, and releases for Mythos-class models.
In the same letter, Lutnick said he had determined that safeguards were in place to allow certain trusted partners to access the Claude Mythos 5 Model. That phrasing points to a controlled release rather than a full return to ordinary availability.
Why The Pressure Built
The source describes pressure on the Trump administration to change the case-by-case regulatory environment it had recently adopted. One factor was competition: rival cybersecurity-focused models were improving and, in some cybersecurity-focused benchmarks, pulling ahead.
Another source of pressure came from within the US AI industry. The concern was that China could advance in AI while leading US AI labs were sidelined.
The situation also affected government users. The source notes that top US government departments, including the National Security Agency, had lost access to Mythos 5.
That combination made the access question more than a private dispute between Anthropic and regulators. It affected model availability for companies, cyber defenders, infrastructure providers, and government departments that had been using or seeking access to advanced cybersecurity AI systems.
A Limited Preview, Not A Final Settlement
Anthropic now appears to have a deal similar to OpenAI’s: a limited preview for approved organizations, including trusted enterprises and the US government itself. Both companies are hoping for broader availability, including enterprise deals and public access.
But the next step remains under the Trump administration’s control. The source says many have called for AI regulation, including AI labs, but some tech leaders do not see the past two weeks as the right model for how that regulation should work.
OpenAI made that point directly in its GPT-5.6 blog post, saying it does not believe this kind of government access process should become the long-term default. The company said it was taking the short-term step because it saw it as the strongest path toward broader availability in the coming weeks while working with the Administration on a cyber Executive Order framework and a repeatable process for future model releases.
For Anthropic, the immediate outcome is progress, but not closure. Mythos 5 is returning to a limited set of users. Fable 5 remains unavailable for general use. And Lutnick’s letter says that all other requirements from the June 12 letter remain in effect until further notice.
Lutnick also wrote that he reserves the right to reevaluate and adjust the scope of license requirements on Mythos 5 and Fable 5 if circumstances change. That leaves the current arrangement open to revision, expansion, or further restriction.