Claude Fable 5 Shutdown and Restoration: US Export-Control Timeline and Current Models

💡 TL;DR / Summary (BLUF)

• What happened: On June 12, 2026, a US export-control directive forced Anthropic to suspend Claude Fable 5 and Mythos 5. Unable to separate users by nationality in real time, Anthropic disabled both models for every customer worldwide.
• Resolution: The Department of Commerce lifted the controls on June 30, and Anthropic restored Fable 5 on July 1, 19 days after the suspension. Mythos 5 came back only for a set of approved US organizations.
• What to use now (September 2026): Fable 5.1 (shipped September 1, $10/$50 per million tokens) for the hardest work, or Opus 5.5 ($4/$20) as the cheaper heavy-coding default. Leaving your model on default is still the safest setting.

“The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.” — Anthropic, official statement, June 12, 2026

Update: Access Restored (September 2026)

This post was first published on June 13, in the middle of the suspension. The situation has since resolved, so the sections below now separate what happened then from what applies now.

• June 30 — The US Department of Commerce lifted the export controls on Fable 5 and Mythos 5.
• July 1 — Anthropic restored access to Fable 5 on Claude Platform, Claude.ai, Claude Code and Cowork. Access through AWS, Google Cloud and Microsoft Foundry followed afterwards. Anthropic also deployed an improved safety classifier that it says blocks the reported bypass technique “in over 99% of cases.”
• Mythos 5 — Restored only for a set of US organizations, following US government approval on June 26. It remains limited-availability through Project Glasswing.
• Subscription plans — On Pro, Max, Team and select Enterprise plans, Fable 5 was usable at up to 50% of weekly usage limits through July 7, after which Anthropic moved it to usage credits.
• September 1 — Anthropic shipped Fable 5.1 and Mythos 5.1 at the same $10 input / $50 output per million tokens, with cache reads cut to $0.25 (from $1 on Fable 5). The release notes list 30-day data retention as a requirement for these models.

The original June 13 reporting follows, kept as the record of the event.

What Happened on June 12 (Original Report)

If you tried to use Claude Fable 5 today and the app told you it was “currently unavailable,” you weren’t imagining it. Three days after launching its most capable model, Anthropic abruptly pulled Fable 5 and Mythos 5 for all users — not because of an outage or a pricing change, but because of a US government directive. I hit it mid-session myself: selecting Fable 5 now drops straight to Opus 4.8.

One clarification up front, because the word “banned” is doing a lot of work in search results: this is not a permanent removal or Anthropic’s choice. It is a government export-control suspension that Anthropic is complying with while openly disputing, and the company says it is trying to restore access. Everything below is sourced from Anthropic’s own statement and major outlets; where something is unconfirmed or speculative, it is labeled as such.

Timeline: From Launch to Restoration

• June 9 — Anthropic launches Fable 5 and Mythos 5, calling Fable 5 the most powerful generally available model it has shipped. (TechCrunch)
• June 10 — Early users complain the new guardrails are overcautious, refusing innocuous prompts — “blocked us at hello.” (The Register)
• June 12, 5:21pm ET — Anthropic receives a US export-control directive citing national security. About four hours later it disables Fable 5 and Mythos 5 for everyone. (Anthropic, CNBC)
• June 13 — Still suspended at the time of this post’s first publication. Anthropic says it is disputing the basis and working to restore access. (Bloomberg)
• June 26 — The US government approves restoring Mythos 5 for a set of US organizations. (Anthropic)
• June 30 — The Department of Commerce lifts the export controls on both models.
• July 1 — Fable 5 access restored for users globally on Anthropic’s own surfaces, 19 days after the suspension. (Anthropic)
• July 7 — The promotional allowance on subscription plans ends; Fable 5 moves to usage credits.
• September 1 — Fable 5.1 and Mythos 5.1 launch. (Claude Platform release notes)

What It Looked Like in the Client (June 12–30)

The clearest evidence is the client itself. Fable 5 still appears in the model picker, but it is greyed out and labeled “Currently unavailable,” and the message area returns “Claude Fable 5 is currently unavailable.” Selecting it lands you on Opus 4.8 instead.

The client states it outright: Fable 5 is currently unavailable, and the available fallbacks are Sonnet 4.6, Haiku 4.5, and the Opus 4.x line.

Pick Fable 5 and the selector reverts to Opus 4.8. (Observed behavior; the cause is the documented suspension.)

This matches Anthropic’s note that requests to Fable 5 on the platform return an error and that you should update integrations to other Claude models. If you have a scheduled task pinned to Fable 5, expect it to fail until you repoint it.

Why It Happened — and Anthropic’s Pushback

Anthropic says the government did not provide specific details, but its understanding is that authorities believe they found a method of bypassing or jailbreaking Fable 5’s safeguards. Anthropic reviewed the demonstration and is unconvinced:

• It describes the disclosed technique as a narrow, minor issue — essentially asking a model to read a specific codebase and fix software flaws — and says the same capability is available in other public models, including OpenAI’s GPT-5.5, and used daily by security defenders.
• It argues that if this standard were applied across the industry, it would halt all new model deployments from every frontier provider.
• Per Axios, a separate company had claimed it jailbroke Mythos, and the administration reportedly pushed Anthropic to delay the launch beforehand.

There is an irony Anthropic addresses directly: its CEO said just two days earlier that the government should have the right to block risky AI models — but, Anthropic argues, only when the action is transparent, fair, and grounded in technical facts, which it says this is not.

On safeguards, Anthropic notes it red-teamed Fable 5 for thousands of hours with US and UK governments, private third parties, and internal teams; engineered universal jailbreaks to be extraordinarily expensive; and added a (controversial) 30-day data-retention policy specifically to catch and fix any narrow jailbreaks quickly.

What to Use Now (September 2026)

The June table below this one is kept for the record. For current decisions, use this one. Prices are list prices per million input / output tokens from Anthropic’s pricing page.

Practical steps now:

1. Keep your model on “default” unless you have a measured reason to pin one. During the suspension, the default setting switched to an available model without any change on the user side.
2. Review anything you repointed in June. If you moved scheduled tasks or integrations from Fable 5 to Opus 4.8 during the suspension, decide deliberately whether they belong on Fable 5.1, Opus 5.5 or Sonnet 5 now.
3. Check your plan’s terms for Fable. On subscription plans, Anthropic moved Fable 5 to usage credits after July 7; confirm how your plan meters it before routing heavy work there.

The June 12 Table (Record)

At the time of the suspension, only Fable 5 and Mythos 5 were affected. Everything else was up.

Practical steps:

1. Leave your model on “default.” That way you always land on the best available model and survive changes like this one automatically.
2. Repoint pinned integrations and scheduled tasks off Fable 5 to Opus 4.8 or default, or they’ll keep erroring.
3. Don’t rebuild your workflow around the outage. Anthropic is trying to restore access; this may be short-lived.

Some users report Opus 4.8 feels like a step down from Fable 5 on the hardest coding tasks — that’s subjective, but worth knowing if your work was tuned to Fable 5.

What This Could Mean (Speculation, Labeled)

A few widely-discussed what-ifs — none confirmed, so treat them as open questions, not facts:

• Identity verification could become a requirement to access top-tier models if nationality has to be enforced per user.
• The precedent worries some builders: if a frontier model can be pulled this fast, could the same pressure reach less-capable closed models or even open-weight releases?
• The timeline was genuinely unknown on June 13. (Outcome: the controls were lifted on June 30 and access returned on July 1.)

Written on June 13, during the suspension:

My take, for what it’s worth: if you run Claude Code day to day, the real lesson here isn’t about export law, it’s that pinning your workflow to whichever model just shipped is now a reliability risk, not just a taste preference — Fable 5 lasted three days before it vanished out from under everyone using it. The jailbreak rationale Anthropic is disputing sounds thin enough (reading a codebase and patching bugs, a capability it says GPT-5.5 already has) that I wouldn’t bet on a fast resolution either way. Until this settles, I’m keeping my own setup on “default” rather than hard-pinning Fable 5 or Opus, and I’d hold off building anything — scheduled tasks, CI agents, whatever — around the assumption that a specific frontier model stays reachable next week.

FAQ

Q. Is Claude Fable 5 available again?

Yes. Anthropic restored Fable 5 on July 1, 2026 on Claude Platform, Claude.ai, Claude Code and Cowork, after the Department of Commerce lifted the export controls on June 30. Fable 5.1 followed on September 1.

Q. Why was Fable 5 blocked for US users as well?

The directive targeted any foreign national, but Anthropic couldn’t reliably distinguish US users from foreign nationals in real time, so it disabled the models for everyone to comply. US users were caught too.

Q. Was the shutdown permanent?

No. It lasted 19 days, from June 12 to July 1. Mythos 5 came back only for a set of approved US organizations and remains limited-availability through Project Glasswing.

Q. Did Anthropic choose to pull the models?

No — it’s a US government export-control directive. Anthropic is complying while publicly disputing the rationale.

Q. Which Claude model should I use now?

As of September 2026: Fable 5.1 ($10/$50 per million tokens) for the hardest work, Opus 5.5 ($4/$20) as a cheaper heavy-coding default, Sonnet 5 ($2/$10) for everyday coding, and Haiku 4.5 ($1/$5) for light tasks. Leaving the model on default is still the safest setting.

Q. What was the jailbreak the government cited?

Anthropic says it amounts to asking a model to read a codebase and fix software flaws — a capability it says is common to other public models (including GPT-5.5) and used by defenders, not a meaningful uplift for misuse.

Sources

• Anthropic — redeploying Claude Fable 5 (access restored): https://www.anthropic.com/news/redeploying-fable-5
• Anthropic — Claude Platform release notes (Fable 5.1, Mythos 5.1): https://platform.claude.com/docs/en/release-notes/overview
• Anthropic — Claude API pricing: https://platform.claude.com/docs/en/about-claude/pricing
• Anthropic — official statement on the suspension: https://www.anthropic.com/news/fable-mythos-access
• CNBC: https://www.cnbc.com/2026/06/12/anthropic-disables-access-to-fable-5-and-mythos-5-to-comply-with-government-directive.html
• Bloomberg: https://www.bloomberg.com/news/articles/2026-06-13/anthropic-says-us-limits-foreign-access-to-fable-5-mythos-5
• NBC News: https://www.nbcnews.com/tech/tech-news/anthropic-suspends-new-ai-models-fable-mythos-government-directive-rcna349901
• The Register (pre-suspension refusal complaints): https://www.theregister.com/ai-and-ml/2026/06/10/anthropic-claude-fable-5-refuses-innocuous-prompts/5253754

Updates & Changelog

• 2026-09-28 — Updated after the restoration. Added the June 26 – September 1 timeline (controls lifted June 30, access restored July 1, usage-credit change July 7, Fable 5.1 on September 1) and a current model table with list prices, all from Anthropic’s own pages. The June 13 reporting is kept as the record.
• 2026-06-13 — Initial publication. Facts compiled from Anthropic’s statement and major outlets; client behavior is first-hand. Status was “suspended, restoration pending” at publication — this is a developing story, so confirm current model availability in your own client before relying on it.

Updated September 28, 2026. The June sections describe the suspension as it happened; the “Update” and “What to Use Now” sections reflect current status. Check your client’s model picker before relying on a specific model.

Educational content only — not investment or financial advice. Data, prices, and tool specifications change; verify independently and paper-trade before risking capital.