Leaked Code Hints at Anthropic’s Potential Revival of Fable 5

Fable 5 could be making a comeback—and when it does, there’s a good chance it won’t cost you anything beyond your regular plan.

A self-described leaker who posts under the handle “synthwavedd” on X revealed on Tuesday that version 2.1.190 of Claude Code contains updated text pointing toward Fable 5’s return, with weekly access folded right into existing subscriptions. The crucial message reads: “You’ve used your Fable 5 usage for this week.”

Originally, Fable 5 was offered to subscribers during a complimentary trial period, after which it was meant to transition to a pay-as-you-go credits system. However, the Trump administration stepped in before that could happen, blocking the model mere days after it launched and claiming it posed a national security risk if ever jailbroken.

The newly discovered text does away with that paywall separation entirely and suggests a recurring weekly quota—imagine it as a bundled data allowance that refreshes every seven days rather than requiring you to dip into your wallet for additional top-ups.

The leaker further pointed out that wording like “purchased separately from your plan” seems to have been stripped from the relevant interface text, hinting that the underlying payment structure is being overhauled—not just cosmetically rephrased.

The exposed text is authentic

Decrypt independently confirmed that these text strings are genuine. Our team fetched the macOS Apple Silicon build of Claude Code directly from npm via npm pack @anthropic-ai/claude-code-darwin-arm64@2.1.190, unpacked the archive, and executed strings -a on the packaged claude binary. The updated Fable 5 message was present: “You’ve used your included Fable 5 usage for this week. Continuing on Fable 5 uses usage credits.”

None of this guarantees Fable 5 will be available right away. Adjustments like these are routine during product development—companies draft interface language for upcoming features well in advance of any public rollout. But production code reveals a team’s true development direction. When text strings shift within a live binary, it’s a clear sign that something is actively being worked on.

That said, some users have noticed that Amazon Bedrock still lists references to Claude Fable 5 among its available models, suggesting this rollout could happen sooner rather than later.

Anthropic’s warming relations with Washington

Fable 5, the company’s first Mythos-class model to be released to the public, was pulled from availability on June 12 following an emergency export control order from the Trump administration that cited a jailbreak vulnerability. Anthropic complied with the order but pushed back on the severity of the finding, contending that any model currently available on the market could exhibit the same vulnerability without requiring any special bypass.

What followed was a very public standoff that escalated quickly. Anthropic CEO Dario Amodei ended up in an awkward seat directly across a large oval table from Trump at the G7 summit in France—OpenAI CEO Sam Altman beside the president, Amodei next to French President Emmanuel Macron.

Anthropic is not a company that currently has much goodwill with Trump. The firm declined to let the Pentagon deploy its AI in certain warfare-related applications, a decision that ultimately resulted in it being designated a supply-chain risk. Trump has also called Anthropic “woke”—a characterization that rarely wins favor with the president.

But the landscape appears to have changed this week. According to Wired, Anthropic replaced Dario Amodei with cofounder Tom Brown as its lead contact in White House talks over Fable 5’s return. The administration has reportedly been more receptive since Brown stepped into the role.

Brown has been meeting with Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross since the middle of June. The discussions now focus on establishing a common framework for assessing AI jailbreak incidents—in essence, agreed-upon standards for determining when a model vulnerability genuinely constitutes a national security threat and when it does not.

Trump himself stated last week that he no longer considers Anthropic a national security threat, reversing a stance he had held as recently as mid-June. Anthropic issued a carefully crafted statement in response, expressing gratitude to the administration for its “ongoing partnership.” The model remains offline as of today, but the political climate has clearly cooled.

If the latest strings in Claude Code are any guide, Anthropic is now building something closer to what subscribers originally anticipated: Fable 5 access on a weekly basis, incorporated into the plan itself rather than being doled out through a separate credits wallet.

The model’s status page at status.claude.com still displays the June 13 suspension as active, with no date announced for when service will be restored.