Anthropic Relaunches Mythos 5

After two weeks of intense negotiations with the Trump administration, Anthropic’s Mythos 5 model has received a partial reprieve, allowing a select group of organizations access. This development comes via a government letter to Anthropic, seen by The Verge. However, Fable 5, the public-facing counterpart in the Mythos-class series, remains unreleased, with no clear timeline for its broader availability.

The letter, dated June 26th and addressed from Commerce Secretary Howard Lutnick to Anthropic co-founder Tom Brown, who spearheaded the recent negotiations, indicates a “revision to the license requirements.” This adjustment follows Anthropic’s collaborative efforts with the U.S. government to “address risks” associated with both the Mythos 5 and Fable 5 models.

In a statement to The Verge, Anthropic spokesperson Danielle Ghiglieri confirmed that the company “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.” Ghiglieri further stated, “We are working to provision the approved set of providers and restore their access to Mythos 5 as quickly as possible. We are pleased to see this progress and continue to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again.”

Crucially, the U.S. government has not rescinded the export control directive imposed two weeks prior, which prohibits foreign nationals, including Anthropic's own staff, from accessing either model. Instead, an exception has been carved out specifically for Mythos 5, granting access to a select group of organizations—a precedent set by the approval of OpenAI’s GPT-5.6 earlier today. This exception, as detailed in the letter, permits both non-U.S. national Anthropic employees and non-U.S. national members of the approved organizations to utilize Mythos 5.

Commerce Secretary Lutnick underscored the positive outcome, writing, “These efforts have yielded significant progress.” He further noted Anthropic’s commitment to "work with the U.S. government on protocols and standards and releases for [Mythos-class models]." Lutnick concluded, “In light of this progress ... I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model.”

This partial approval follows mounting pressure on the Trump administration to reconsider its recently adopted case-by-case regulatory approach. Concerns intensified as Anthropic’s competitors continued to advance their cybersecurity-focused models, in some instances surpassing Mythos 5 on key benchmarks. Within the U.S. AI industry, there were growing anxieties about potential AI advancements by China while leading American AI laboratories remained restricted. Furthermore, critical U.S. government entities, such as the National Security Agency, had also been deprived of access to Mythos 5.

Anthropic now operates under terms similar to those granted to OpenAI: a limited preview release of its model, exclusively for approved organizations, including trusted enterprises and the U.S. government. Both AI developers are optimistic for a swift transition to more general availability, encompassing broader enterprise agreements and public access, such as for Anthropic’s Fable 5. However, this ultimate decision rests with the Trump administration. While calls for AI regulation are widespread, including from AI labs themselves, a prevailing sentiment among some tech leaders suggests that the regulatory method employed over the past two weeks has been counterproductive.

Indeed, OpenAI articulated its concerns in a blog post regarding GPT-5.6, stating, “We don’t believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them. We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks, while we work with the Administration to develop the cyber Executive Order framework and a repeatable process for future model releases.”

In his correspondence to Anthropic, Lutnick reiterated, “All other requirements of the June 12 letter remain in effect until further notice.” He further cautioned, “I reserve the right to reevaluate and adjust the scope of license requirements on [Mythos 5 and Fable 5], should circumstances change.”