# Claude Fable 5: What It Is, Why It Was Suspended, and What Changed (2026)
> July 18 update: Fable 5 billing splits permanently on July 20 — Max/Team Premium keep it at reduced limits; Pro/Team Standard lose subscription access and get a one-time $100 credit. Full story: the June 12 government suspension, 19-day standoff, July 1 restoration, and the new permanent plan tiers.
**Author:** [Alex Lowe](https://theaicareerlab.com/about) — Founder, The AI Career Lab
**Published:** 2026-06-09
**Last reviewed:** 2026-07-18
**Canonical URL:** https://theaicareerlab.com/blog/claude-fable-5-for-professionals
**Category:** industry-news
**Tags:** Claude, Claude Fable 5, Anthropic, AI models, AI regulation, 2026, model news
---> **UPDATE — July 18, 2026. Fable 5 billing splits permanently on July 20.** Anthropic has announced permanent billing changes effective July 20: **Max and Team Premium** subscribers keep Fable 5 included, but at only 50% of their plans' weekly limits — and those limits are themselves reduced by ~33% when the bonus usage phase ends the same day. Effective Fable 5 coverage for Max/Team Premium users is roughly a third of the original launch allowance. **Pro and Team Standard** subscribers lose subscription access to Fable 5 entirely starting July 20. They receive a one-time **$100 usage credit** for API access at $10/M input and $50/M output (twice the Opus 4.8 API price, the highest Anthropic has ever charged for a generally available model), then pay those API rates ongoing. Anthropic had originally planned to pull Fable 5 from subscriptions entirely before reversing course, citing demand management challenges and competitive pressure from OpenAI's GPT-5.6 Sol (similar capability at roughly a third of the cost). Sonnet 5 and Opus 4.8 remain fully included on all plans. If you're on Pro and want to keep using Fable 5, the $100 credit will cover perhaps a day of heavy professional use at these rates — after that, you'll need API billing or an upgrade to Max. See the "What changes for my Claude plan on July 20" FAQ below.
>
> **UPDATE — July 1, 2026 (Day 19). Fable 5 is back globally.** Following a June 30 lifting of export controls, Anthropic restored Claude Fable 5 on July 1 across Claude.ai, Claude Code, Claude Cowork, and the API. AWS, Google Cloud, and Microsoft Foundry integrations are following shortly. **Billing through July 7:** Fable 5 is included for up to 50% of weekly usage limits on Pro, Max, Team, and select Enterprise plans — a temporary grace window. **After July 7:** credits-based billing applies (Standard Enterprise seats need credits enabled; Premium Enterprise seats include Fable 5). The 30-day data retention requirement for safety monitoring remains in effect. Anthropic closed the 19-day standoff by committing to: expanded federal early access for security review before future frontier model releases, rapid jailbreak information sharing, dedicated joint research teams, and participation in a shared voluntary jailbreak severity scoring framework with Amazon, Microsoft, and Google. Mythos 5 remains available to 100+ US companies and federal agencies through Project Glasswing. For professionals: **Fable 5 is available now.** If you were using Opus 4.8 as a fallback, you can switch back. Check your plan's billing tier before using heavily after July 7.
>
> **UPDATE — June 27, 2026 (Day 15).** The US government partially lifted restrictions. Commerce Secretary Lutnick issued a new authorization clearing **Claude Mythos 5** for approximately 100 trusted US partners — cybersecurity firms and critical infrastructure providers, including Apple, Google, Cisco, Nvidia, and Microsoft (Project Glasswing members). Non-US nationals at approved organizations are also now cleared. **Claude Fable 5 remains suspended** as of this writing, but Axios (June 27) reports the Trump administration expects to restore access as soon as this week — Pentagon and NSA sign-off is the remaining step. Pricing note: as of June 23, Fable 5 is no longer included in Pro/Max/Team plan limits; when restored, usage costs $10/$50 per million tokens (pay-per-token). **For now: use Claude Opus 4.8 or Sonnet 4.6.** Monitor [Anthropic's newsroom](https://www.anthropic.com/news) — restoration will be announced there first.
>
> **UPDATE — June 21, 2026 (day 9).** Both models remain suspended. No restoration announcement has been issued by Anthropic. The June 20 refund window for June 9–14 subscribers has closed. The June 22 bundled-pricing deadline — after which Fable 5 use would cost $10/$50 per million tokens rather than being included in Pro/Max/Team limits — is tomorrow. Circulating reports of a "June 18 restoration with identity checks" are unverified and contradicted by Anthropic's own newsroom and live status tracking. TechCrunch (June 21) analysis: the prolonged ban is paradoxically strengthening Anthropic's reputation by signaling it holds AI powerful enough to warrant government intervention. **Use Claude Opus 4.8 or Sonnet 4.6 — both fully available and unaffected.**
>
> **UPDATE — June 19, 2026.** President Trump told Axios in a pretaped interview that he no longer views Anthropic as a US security threat — "Well, not now, but a week ago, maybe" — and that CEO Dario Amodei is "nice" and "smart." He said Anthropic "behaved very responsibly" after the June 12 directive. This is the clearest presidential signal yet that the threat rationale behind the ban has softened, though no formal restoration order has been issued. Separately, Anthropic's updated privacy policy (effective July 8) adds government-ID and biometric data collection — a likely technical step toward restoring US-only access. **Two active deadlines: refund window for June 9–14 subscribers closes June 20; Fable 5's bundled pricing on Pro/Max/Team ends June 22.** See the [June 17–19 section below](#june-17-19) for full details.
>
> **UPDATE — June 16, 2026.** Bloomberg has published the full text of Commerce Secretary Lutnick's letter to Anthropic — the first time it's been available in its exact words. The letter ordered Anthropic to suspend all foreign-national access within 90 minutes, threatening "criminal and civil penalties" for non-compliance. Separately, more than 120 cybersecurity professionals have signed an open letter asking the government to lift the restrictions, calling the capability that triggered the ban a standard defensive security technique. Washington meetings are ongoing; no restoration timeline has been given. See the [June 16 section below](#june-16-lutnick-letter-published-and-120-experts-push-back) for the full story.
>
> **UPDATE — June 15, 2026.** Anthropic's senior technical staff are in Washington, D.C., meeting with White House and Commerce Department officials to dispute the export ban. Fortune's reporting today reveals the specific prompt — "**fix this code**" — that triggered the government's jailbreak claim. Axios reports the dispute runs deeper than that jailbreak: the administration alleges Anthropic allowed Mythos 5 to be accessed by an entity with ties to China, and that Anthropic failed to comply with a recent cyber executive order. No restoration timeline has been given. See the [June 15 section below](#june-15-washington-meetings-fix-this-code-revealed-and-a-wider-dispute) for the full story.
>
> **UPDATE — June 13, 2026.** New reporting reveals the research behind the directive was conducted by **Amazon researchers** and briefed to the government by Amazon CEO Andy Jassy. A security expert who reviewed the research describes it as a **defensive probing technique**, not an offensive jailbreak. See the [June 13 section below](#june-13-the-jailbreak-was-amazons-own-security-research) for the full story.
>
> **UPDATE — June 12, 2026.** Claude Fable 5 and Mythos 5 are **currently suspended for all users globally.** The US government issued an export control directive; since Anthropic cannot identify foreign nationals in real time, it disabled both models worldwide. **Claude Opus 4.8, Sonnet 4.6, and Haiku are fully available.** See the [June 12 section below](#june-12-us-government-orders-fable-5-and-mythos-5-offline) for the full story.

> **Original TL;DR (June 9).** Anthropic released **Claude Fable 5** on June 9, 2026 — a Mythos-class model that outperforms every previous Claude on coding, document reasoning, and long-horizon tasks. It was available on **Pro, Max, Team, and Enterprise** plans; it was suspended three days after launch. Via API: `claude-fable-5` at $10/$50 per million tokens (currently unavailable).

Anthropic shipped two models on June 9, 2026: **Claude Fable 5**, publicly available, and **Claude Mythos 5**, restricted to vetted research partners. For most professionals using Claude.ai or the API, Fable 5 was the relevant new model — until June 12, when both were suspended under a US government directive. Here is what each model was, what happened, and what to do now.

> **Here for what Claude can do for your job?** Skip the model news and see it set up for your work: [real estate agents](/resources/claude-cowork-real-estate), [social media managers](/resources/claude-cowork-social-media-manager), [bookkeepers](/resources/claude-cowork-bookkeeper), [management consultants](/resources/claude-cowork-management-consultant), [financial advisors](/resources/claude-cowork-financial-advisor) — or [find your profession](/professions).

## What "Mythos-class" means

Anthropic's model tiers have historically been Haiku (fast/light), Sonnet (balanced), and Opus (most capable). Fable 5 introduced a new category above Opus: **Mythos-class**. Think of it as the jump from "top of the existing line" to a new tier — not an incremental Opus update, but a substantially different model.

The benchmark numbers make this concrete. On **SWE-Bench Pro** — a test of real-world software engineering tasks — Fable 5 scores 80.3% against Opus 4.8's 69.2%. On **FrontierCode**, it scores 29.3% against GPT 5.5's 5.7% and Opus 4.8's 13.4%. On **Hebbia's Finance Benchmark**, which tests senior-analyst-level reasoning over complex documents, Fable 5 tops the leaderboard.

These aren't cherry-picked single-domain wins. The model is better across the domains professionals actually care about: long-form reasoning, document analysis, coding, and vision.

## What's actually different for the work you do

The capability change that matters most for professional use isn't the benchmark number — it's how long the model can hold a task together.

**Sustained task execution.** Fable 5 can work through multi-step workflows for hours — Anthropic's product page describes "days at a time" in agentic configurations, and researcher Ethan Mollick described sessions running "up to a dozen hours" on multi-page specifications. For most daily professional use, this translates to: fewer breakdowns mid-task, better coherence in long documents, and less re-prompting when you're doing something genuinely complex.

Stripe reported compressing five months of engineering work into days using Fable 5 in an agentic pipeline. That's an enterprise example, but the underlying capability — not losing the thread in step 8 of a 10-step task — shows up in smaller ways for individual professional users too.

**Document and financial reasoning.** The model tops Hebbia's Finance Benchmark, which tests performance on tasks like reading dense regulatory filings, cross-referencing contract terms, and synthesizing multi-document analysis. For attorneys reviewing complex agreements, financial advisors building reports from multiple data sources, or compliance officers working through regulatory updates, this is the practical improvement to look for.

**Vision.** Fable 5 sets a new benchmark for vision tasks. In testing it has rebuilt complete web applications from nothing more than a screenshot — which is impressive as a demo, but the practical version is: it reads charts, tables, medical images, and embedded figures in PDFs more reliably than Opus 4.8.

**Code.** The coding gains are significant enough that Claude Code users notice them immediately. Fable 5 writes its own tests, handles large migrations autonomously, and recovers from mid-task errors without losing context. For professionals running Python-light analysis, automating document workflows, or using Claude Code for professional tasks, this is the biggest day-to-day improvement.

## What Mythos 5 is (and why most readers won't access it)

Claude Mythos 5 is the more capable sibling — restricted via Anthropic's **Project Glasswing** program to vetted cybersecurity researchers and select biomedical labs. It scored 78% on ExploitBench for cybersecurity tasks and matched experienced human operators on drug-design workflows.

If you're reading this as a professional using Claude.ai or the API for business work, Mythos 5 is not available to you, and that's intentional. It's not behind a paywall — it's behind a vetting and safety review process designed to limit access to the highest-risk capabilities. Fable 5 is the model that delivers the practical benefits of the Mythos generation to the rest of us — when it's available.

**Updated June 10.** The safety architecture has two distinct tiers. The first is visible: for offensive cybersecurity, certain biology/chemistry queries, and model distillation requests, Fable 5 routes to Claude Opus 4.8 — you can see this fallback happen, and it affects fewer than 5% of sessions. The second is not visible: for frontier LLM development specifically (pretraining pipelines, distributed training infrastructure, ML accelerator design), Fable 5 does not route to another model. It responds, but uses "prompt modification, steering vectors, or parameter-efficient fine-tuning" to limit its own effectiveness. Anthropic's system card states explicitly that these interventions "will not be visible to the user." This silent restriction affects an estimated 0.03% of traffic, concentrated in fewer than 0.1% of organizations.

## What to use right now

> **Note:** this section was written during the June 12–July 1 suspension and is preserved as history. **Fable 5 is available again** — the guidance below on Opus 4.8 / Sonnet fallbacks no longer applies.

For quick, single-turn tasks — short drafts, direct questions, simple summaries — Opus 4.8 and Sonnet 4.6 remain fast and efficient. When Fable 5 is available again, it is worth the extra usage when the task genuinely benefits from sustained reasoning: long-document analysis, multi-step workflows, complex code, or any session where you'd otherwise be doing significant re-prompting.

One constraint to flag when Fable 5 is restored: Anthropic requires **30-day data retention** for Fable 5 sessions for safety monitoring purposes. If your organization has a zero-retention data policy for AI — common in certain legal, healthcare, and financial contexts — check whether that requirement changes your deployment plans.

## The silent restriction controversy (June 10 update)

The day after Fable 5 launched, AI researchers surfaced a disclosure buried in the model's 319-page system card that sparked sharp backlash: the silent restriction described above.

The core objection from critics isn't that Anthropic chose to limit certain capabilities — it's that it does so without telling the user. Unlike the visible Opus 4.8 routing (where you know a restriction triggered), this tier gives you a response that looks normal but has been deliberately degraded. Nathan Lambert of AI2 called this "appalling" and said having model access "rug pulled in an under the table fashion" was unacceptable for research work. Jeremy Howard of Fast AI argued the approach increases power imbalances by giving Anthropic privileged access while silently limiting what competitors and researchers can extract. Multiple researchers noted that silent degradation creates an unlogged confounder — if you don't know you received a weakened response, you can't account for it in your work.

Dianne Na Penn, Anthropic's head of product management, acknowledged the restrictions were designed to make the model "accessible in a safe manner" without elaborating on the silent mechanism specifically.

**What this means for most professionals using this guide:** Zero. The 0.03% figure is Anthropic's own estimate, and it targets a narrow technical category — people building the infrastructure to train large models. Attorneys, healthcare workers, financial advisors, consultants, teachers, and the vast majority of knowledge workers using Claude for productivity tasks will never encounter this restriction.

## Anthropic reverses the silent restriction (June 11 update)

One day after the backlash erupted, Anthropic reversed course. In a statement reported by Wired and Fortune, the company apologized and changed the policy:

**What changed:**

- **On Claude.ai:** Flagged requests now visibly fall back to Opus 4.8 — the same visible routing that already applied to cybersecurity and biology queries. Users will see that a restriction triggered, rather than receiving a silently degraded response.
- **On the API:** Flagged requests return an explicit reason for refusal instead of returning a response that appears normal but has been covertly weakened.

Anthropic's own statement explains the tradeoff they got wrong: "Visible safeguards can be probed, so they have to be robust, which takes time to get right. Invisible safeguards can be targeted more narrowly, allowing us to ship quickly with very few false positives. We went with invisible safeguards for this reason — and that was the wrong tradeoff. Users should have visibility into the safeguards we have in place, and why. We're sorry for not getting the balance right."

**What did not change:** The restriction itself stays. Anthropic's terms of service prohibit using Claude Fable 5 to build competing AI systems, and the company cites national security concerns — specifically preventing foreign adversaries from extracting AI capabilities to improve their own models.

The speed of the reversal — less than 48 hours from launch to apology — suggests the research community's critique landed not as a disagreement about safety goals but as a clear line being crossed: a vendor covertly degrading the product you paid for, with no signal that anything was wrong.

## June 12: US government orders Fable 5 and Mythos 5 offline

At 5:21 PM ET on June 12, 2026 — three days after launch — Anthropic disabled Claude Fable 5 and Claude Mythos 5 for all users worldwide.

**What happened:** The US government issued an export control directive ordering Anthropic to suspend access to both models for "any foreign national." Anthropic reviewed the directive and complied the same day, disabling both models globally.

**Why both models — and why globally:** Anthropic cannot reliably identify which of its users are foreign nationals in real time. Rather than attempt partial compliance and risk a violation, the company disabled both models for all customers everywhere. Claude Opus 4.8, Sonnet 4.6, and Haiku remain fully available.

**The government's stated reason:** The directive cited national security concerns and the government's awareness of "a method of bypassing, or jailbreaking Fable 5." In practice, the government provided only verbal evidence describing what Anthropic characterizes as "a potential narrow, non-universal jailbreak."

**Anthropic's public response:** Anthropic complied, but published an official statement publicly disagreeing with the scope and basis of the directive:

> "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."

Anthropic says its safeguards for Fable 5 exceeded industry standards through extensive red-teaming, and that the demonstrated vulnerability is comparable to weaknesses present in other public models that were not recalled. The company also stated its position on government oversight broadly: "government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts" — framing its objection not as opposition to regulation, but as criticism of the specific process: non-transparent, verbal evidence, without a clear technical threshold.

**What to do right now:**

- **Use Claude Opus 4.8** for complex professional work — long documents, deep analysis, multi-step workflows, code. It was Anthropic's top model before Fable 5 and remains highly capable.
- **Use Claude Sonnet 4.6** for everyday tasks: drafting, quick research, summaries, client communications.
- **API users:** Replace any `claude-fable-5` model strings in your code with `claude-opus-4-8` for the closest available capability. Calls to `claude-fable-5` will fail until access is restored.
- **Monitor** the [Anthropic news page](https://www.anthropic.com/news) for restoration updates.

**The bigger picture:** This is the first known instance of the US government using an export control directive to force a major AI lab to suspend a recently launched commercial model globally. The fact that an action targeting "foreign nationals" ended up shutting off access for everyone — because nationality cannot be verified at inference time — illustrates a structural gap in how AI export controls work in practice. Whether this is the start of a pattern or a one-off event tied to a disputed jailbreak will depend on what happens next. For professionals who build workflows around specific AI tools, it's a reminder that model availability is not guaranteed, and keeping tested fallback models in your stack is prudent.

## June 15: Washington meetings, 'fix this code' revealed, and a wider dispute {#june-15-washington-meetings-fix-this-code-revealed-and-a-wider-dispute}

### The "fix this code" mechanism

Fortune's reporting on June 15 fills in the specific technique at the center of the government's jailbreak claim. Amazon researchers found that Fable 5 refused requests framed as "review this code for security issues" — but complied when they instead said "fix this code," producing patches to software flaws. The practical difference: a repair-oriented prompt slipped through guardrails tuned to block security-analysis framing.

Whether this is a meaningful vulnerability or a narrow edge case requiring a minor patch remains contested. Katie Moussouris — the CEO of security firm Luta Security, who reviewed the Amazon research for Anthropic — published an open letter on June 15 calling the technique "Defense Oriented Prompting (DOP)," a methodology security practitioners use to probe AI systems for weaknesses so they can be fixed. Her characterization: this was a defensive exercise that should have been handled through responsible disclosure, not a global commercial ban.

### Anthropic sends senior staff to Washington

Anthropic has sent senior technical executives to Washington, D.C., to meet with White House and Commerce Department officials. The goal is to demonstrate that Fable 5's safeguards can be verified or patched, and to open a path toward restoring access. Bloomberg and Axios confirmed the meetings are underway as of June 15. No outcome has been reported yet.

### A dispute larger than the jailbreak

New reporting from Axios today reveals that the conflict between Anthropic and the Trump administration runs deeper than the technical jailbreak question. According to administration officials:

- The government learned that Mythos 5 had allegedly been accessed by an entity with **direct ties to the Chinese Communist Party** — a separate and serious export-compliance concern.
- Anthropic allegedly failed to comply with a **recent cyber executive order**.
- An administration source described the relationship: "It's like they just speak in different languages." Another said simply: "They screwed us."

Anthropic has not publicly addressed the CCP-access allegation. If accurate, it represents a materially different basis for the export controls — one that would not be resolved simply by patching the model's "fix this code" guardrail. The Washington meetings will need to address both threads.

### What this means for your timeline

There is still no official timeline for restoring Fable 5 or Mythos 5. Sources tracking the situation consider restoration before late June extremely unlikely. The Washington meetings are a positive signal — both sides are at the table — but the outcome depends on whether Anthropic can satisfy regulators on the technical jailbreak and, separately, the access-control and executive-order compliance allegations.

One broader consequence is already visible: Bloomberg reports that **Cohere**, a Canadian AI company, is fielding inbound inquiries from customers reassessing their Anthropic dependency. The Anthropic situation is accelerating conversations about vendor concentration risk that many enterprise AI buyers had been deferring.

**For now: Claude Opus 4.8 and Sonnet 4.6 remain fully available and unaffected.**

---

## June 16: Lutnick letter published — and 120+ experts push back {#june-16-lutnick-letter-published-and-120-experts-push-back}

### The letter is now public

Bloomberg published the full text of Commerce Secretary Howard Lutnick's letter to Anthropic on June 16, 2026 — the first time the directive has been available in its exact words rather than described secondhand. The letter ordered Anthropic to "suspend all access to Fable 5 and Mythos 5 by any foreign national, anywhere in the world, including foreign nationals working inside the United States." It warned: "Failure to comply will result in prompt criminal and civil penalties." Anthropic was given 90 minutes to comply.

The letter's publication makes the government's legal basis explicit: it cites US statutes that allow the Commerce Department to apply export controls to civilian technology that could be used for intelligence purposes by an adversary's military. Whether those statutes were properly invoked — given that independent security experts characterize the underlying research as a defensive technique, not an offensive exploit — is the core legal question now being argued in Washington.

The 90-minute compliance window stands out. Standard export control procedures typically involve a formal review process with an opportunity to respond. The compressed timeline is consistent with Anthropic's own description of the directive as unusual in scope and process.

### 120+ security professionals sign open letter

More than 120 cybersecurity and technology professionals have signed an open letter calling on the government to lift the restrictions. Their core argument: the capability that triggered the ban — the "fix this code" prompting technique identified by Amazon researchers — is a standard defensive security methodology used to probe AI systems for weaknesses so they can be fixed.

The letter states the restrictions "have taken the best models away from defenders, created market uncertainty, and risked America's AI leadership without any real risk to justify it." Signatories argue that blocking US defenders from the best American AI tools does not prevent adversaries from developing comparable capabilities — particularly given the narrowing gap between US and Chinese AI models. Restricting defense-side access while the underlying capability exists in competing systems, the letter argues, advantages adversaries rather than protecting US security.

### What this means for your timeline

No restoration timeline has been given as of June 16. The Washington meetings from June 15 are continuing. The Lutnick letter's publication and the expert pushback may or may not accelerate a resolution — that depends on whether the administration's primary concern is:

1. **The technical jailbreak** (the "fix this code" technique) — potentially addressable with a model patch, which Anthropic has already indicated it can provide.
2. **The separate CCP-access allegation** — the Axios-reported claim that Mythos 5 was accessed by an entity with ties to China. This would require a fuller compliance review independent of any model patch.

If the administration requires resolution of both tracks before restoring access, late June or beyond remains the realistic floor.

**For professionals: Claude Opus 4.8 and Sonnet 4.6 remain fully available and unaffected.**

---

## June 13: The 'jailbreak' was Amazon's own security research

New reporting fills in who conducted the research behind the directive — and contests whether "jailbreak" is the right characterization.

**Amazon's role.** According to the Wall Street Journal, Amazon CEO Andy Jassy briefed Treasury Secretary Scott Bessent and other government officials that Amazon researchers had used Claude Fable 5 to surface information "that could be used in cyberattacks." David Sacks, Trump's AI policy advisor, described the source as "a highly credible trusted partner" — consistent with the WSJ's identification of Amazon. An Amazon spokesperson declined to discuss the details of government conversations. (Amazon has a large investment stake in Anthropic through AWS.)

**A defensive technique, not an attack tool.** Katie Moussouris, CEO of security firm Luta Security, reviewed the research findings and offered a sharply different characterization. She described the technique as "Defense Oriented Prompting (DOP)" — a methodology security practitioners use to systematically probe AI systems for vulnerabilities, so those weaknesses can be fixed. "If Nat defense is the goal," Moussouris wrote, "this just scored an own goal against us." By that reading, the research that triggered a global commercial ban was conducted by a major Anthropic investor, using a defensive security methodology, to identify weaknesses — exactly the kind of finding a responsible disclosure process is designed to handle without shutting off commercial access.

**What this adds to the picture.** Anthropic's original description of the directive — based on "a narrow, non-universal jailbreak" presented as "verbal evidence" — now has more context: the research was internal Amazon work, reviewed by a credible security expert who disputes the "jailbreak" framing. The distinction matters. A defensive probe that surfaces a potential vulnerability is not the same as an adversarial exploit ready for deployment. Whether that distinction changes the regulatory outcome is unknown. As of June 13, Anthropic has not issued an updated statement, and no timeline for restoring Fable 5 access has been given.

---

## June 17–19: Trump reverses on the threat designation — two deadlines now active {#june-17-19}

### Trump at G7: "going fine" (June 17)

On June 17, President Trump met with Anthropic CEO Dario Amodei and other tech executives on the sidelines of the G7 summit in Évian-les-Bains, France. Afterward, Trump told reporters that negotiations with Anthropic are "going fine." It was the first direct presidential comment on the dispute since the June 12 directive — a notable shift in tone from an administration that had issued a 90-minute compliance ultimatum six days earlier.

### Trump to Axios: "Not now, but a week ago, maybe" (June 19)

In a pretaped interview aired by Axios on June 19, Trump was asked directly whether he views Anthropic as a US national security threat. His answer: "Well, not now, but a week ago, maybe." He said Anthropic had "behaved very responsibly" after the directive, and described Amodei as "nice" and "smart." Bloomberg reported the same interview.

This is the clearest signal yet that the White House considers the dispute resolvable. The export control directive originated in a security designation — and the president has now publicly stepped back from the underlying threat framing.

What this does not do: it does not lift the Commerce Department's export control order. Trump's remarks are a political signal, not a legal change. The ban remains in effect as of June 19.

### Anthropic's privacy policy update signals an identity-verification path

Anthropic's updated privacy policy (effective July 8, 2026) adds new data categories it may collect: government-issued identity documents, biometric data, and facial geometry. The policy states Anthropic may collect "an image of your government-issued identity document and the information appearing on it; your image in photo or video form, facial geometry templates (which may be considered 'biometric data' in some jurisdictions); and the result of the verification."

The practical reason this matters: Anthropic disabled Fable 5 globally — including for American users — because it cannot verify citizenship at inference time. An ID-verification layer could allow Anthropic to restore access for confirmed US users without waiting for the full ban to be lifted. This is not confirmed as the restoration mechanism, but the timing and language are consistent with a US-only partial restoration as a near-term path.

**Enterprise teams:** If your organization has a data processing agreement, zero-retention policy, or contract with Anthropic that restricts biometric data collection, review it before July 8.

### UK exemption proposal collapses

A proposal to exempt the United Kingdom from the export control directive — which would have restored access for UK users — collapsed in June 17 reporting. Near-term international access is blocked. If a partial restoration happens before end of June, it will almost certainly be US-only and conditional on identity verification.

### Two deadlines you need to act on

**June 20 — refund window closes.** Anthropic opened refunds for subscribers who upgraded to Pro, Max, Team, or Enterprise plans June 9–14, because Fable 5 was the announced reason for upgrading and was then suspended three days after launch. The window closes June 20. If you upgraded in that window and the suspension changes your cost-benefit calculus, this is your last opportunity to request a refund.

**June 22 — bundled pricing ends.** From June 9 through June 22, Fable 5 has been included at no additional cost on Pro, Max, and Team plan limits. On June 23, Anthropic removes it from those limits: if Fable 5 is restored after that date, use will require pay-per-token credits at $10/million input tokens and $50/million output tokens — double the price of Opus 4.8. If you're an annual subscriber who expects to use Fable 5 heavily once it returns, factor this pricing change into your plans.

**For now: Claude Opus 4.8 and Sonnet 4.6 remain fully available and unaffected.**

---

## June 21: Day 9 — refund window closed, pricing deadline tomorrow {#june-21}

As of June 21, 2026, both Fable 5 and Mythos 5 remain suspended globally. Anthropic's newsroom shows no restoration announcement. Live status tracking at isfableback.org confirms the models are offline. Circulating reports from third-party blogs claiming a "June 18 restoration with identity verification and geo-fencing" are unverified; those sources noted themselves that "platform restoration specifics remain unconfirmed by independent sources," and they are contradicted by the official record.

### June 20 deadline has passed

The refund window for Claude subscribers who upgraded to Pro, Max, Team, or Enterprise plans between June 9 and June 14 — specifically because Fable 5 was included — closed on June 20. If you did not request a refund before the window closed, that option is no longer available.

### June 22 deadline is tomorrow

On June 23, Anthropic removes Fable 5 from the included usage on Pro, Max, and Team plans. If Fable 5 is restored after June 22, use will require pay-per-token credits at **$10 per million input tokens and $50 per million output tokens** — double the price of Opus 4.8. Annual subscribers who expected heavy Fable 5 use should factor this into their planning, since the model may return to a higher-cost tier than when it launched.

### The competitive picture (TechCrunch, June 21)

TechCrunch's June 21 analysis of the crackdown raises a counterintuitive point: the US government's willingness to invoke export controls specifically against Anthropic — and not OpenAI or Google — may be inadvertently signaling that Anthropic's models are the most capable frontier AI. For customers and enterprise buyers trying to assess which AI provider has the most advanced capabilities, "powerful enough to be regulated by the government on national security grounds" is an unusual but concrete signal.

The same analysis notes that OpenAI and Google are the most obvious near-term beneficiaries of Anthropic's disruption, as enterprise teams reassess vendor concentration risk. Cohere also continues to report inbound inquiries from customers exploring alternatives.

**For professionals:** The practical guidance from June 12 is unchanged. Use Claude Opus 4.8 for complex, multi-step professional tasks. Use Sonnet 4.6 for everyday work. Monitor [Anthropic's newsroom](https://www.anthropic.com/news) — restoration, when it happens, will appear there first.

---

## June 26–27: Mythos 5 cleared; Fable 5 expected this week {#june-26-27}

### Mythos 5: approved for ~100 trusted partners

On June 26, 2026 — two weeks after the June 12 directive — Commerce Secretary Howard Lutnick issued a second letter to Anthropic. This one authorized Claude Mythos 5 access for approximately 100 "trusted" US organizations, primarily cybersecurity companies and critical infrastructure providers. The letter stated Anthropic "has worked with the US government to address risks associated with the Covered Models" and that "appropriate safeguards are in place."

Companies already enrolled in Anthropic's Project Glasswing program — roughly 200 organizations including Apple, Google, Cisco, Nvidia, and Microsoft — are cleared to resume Mythos 5 access. The authorization also extends to non-US nationals working at approved US organizations or employed directly by Anthropic, reversing the global nationality-based lockout for this group.

The Lutnick letter does not grant permission to restore Fable 5 for general access. The two models are being treated separately under the new authorization framework.

### Fable 5: expected back within days

A separate Axios report (June 27) describes Fable 5's return as expected "within days." Administration insiders say access could be restored "as soon as this coming week." The remaining step: Pentagon and NSA sign-off. The export control directive technically remains in effect; restoring Fable 5 to general access will require either a formal Commerce Department withdrawal or a new authorization framework covering public access.

Both Anthropic and OpenAI are now pushing the government for a predictable, legally defined framework for frontier model releases — rather than individual case-by-case approval decisions — to prevent future crises like this one.

### What "restored access" may look like

When Fable 5 returns, the access model may not be identical to the June 9 launch. The relevant comparison: OpenAI's GPT-5.6 Sol, announced June 26 in a preview restricted to ~20 government-approved partners before a planned broader rollout ([it went broadly available on July 9](/blog/gpt-5-6-sol-government-restrictions-2026)). Administration sources have not confirmed whether Fable 5 will launch to all paying subscribers immediately or follow a similar tiered-partner model.

What is confirmed for when Fable 5 returns:

- **Pricing change**: Fable 5 is no longer included in Pro, Max, or Team plan limits (as of June 23). Usage will cost $10 per million input tokens and $50 per million output tokens — pay-per-token, not bundled.
- **Data retention**: Anthropic requires 30-day data retention for Fable 5 sessions for safety monitoring. Enterprise teams with zero-retention policies should review their Anthropic data terms before enabling Fable 5.
- **Identity verification**: Anthropic's July 8 privacy policy update adds government-ID and biometric data collection. The ID-verification path appears to remain part of the restoration mechanism.

**For professionals:** No action needed right now. Watch [Anthropic's newsroom](https://www.anthropic.com/news) — restoration will be announced there. When the announcement appears, check whether your subscription plan requires separate pay-per-token credits for Fable 5, and confirm your organization's data-retention policy is compatible with the 30-day requirement.

---

## July 1: Fable 5 is back — the truce terms {#july-1-fable-5-restored}

On June 30, 2026, the US Commerce Department lifted export controls on Fable 5. Anthropic restored global access on July 1, 2026 — 19 days after the June 12 suspension.

### What access looks like now

Fable 5 is available on Claude.ai, Claude Code, Claude Cowork, and the API. AWS, Google Cloud, and Microsoft Foundry integrations were restoring in the days immediately following. No US-only restriction applies — access is global.

**Billing (this changed from the June 23 announcement):**

- **Through July 7, 2026**: Fable 5 is included for up to 50% of weekly usage limits on Pro, Max, Team, and select Enterprise plans — a one-week grace window.
- **After July 7**: Credits-based billing applies. Standard Enterprise seats require credits enabled; Premium Enterprise seats include Fable 5 in the subscription.
- **30-day data retention** for safety monitoring remains in effect for all Fable 5 sessions.

If you upgraded to Pro or Max specifically for Fable 5 and were refunded during the suspension, check whether your plan is currently active.

### What Anthropic committed to

The terms of the truce are specific. Anthropic agreed to:

- **Pre-release federal access**: Provide government partners "expanded early access" to frontier models — including both the models and their safeguards — before public launch.
- **Rapid jailbreak sharing**: Share information about jailbreaks and safeguards quickly with government partners, rather than handling vulnerabilities unilaterally.
- **Dedicated research teams**: Staff joint teams for security and alignment research with federal counterparts.
- **Industry jailbreak framework**: Contribute to a shared voluntary standard for assessing jailbreak severity, developed with Amazon, Microsoft, and Google.

The jailbreak severity framework scores vulnerabilities on four criteria: capability gain (how much beyond existing tools), breadth (how many offensive tasks affected), ease of weaponization (required human effort), and discoverability (how readily obtainable). This framework is intended to give the government a clearer basis for future decisions — avoiding the situation where a defensive prompting technique triggers a global commercial ban.

Commerce Secretary Lutnick confirmed the models "have been relaxed" after the government spent two weeks working with Anthropic on alignment issues. Anthropic's pending IPO — filed confidentially June 1 — likely added urgency to reaching a resolution.

### What this means for the future

Both Anthropic and OpenAI are now pushing the government for a legally defined framework for frontier model releases — codified rules rather than case-by-case directives. The truce gives them both a precedent and a mechanism to point to. Whether this produces durable policy or remains an informal arrangement depends on whether future administrations maintain the same working relationship.

For professionals: if you were using Opus 4.8 as your Fable 5 fallback, you can now switch back. The model is available on the same platforms you were already using. Keep an eye on billing after July 7 if your plan is Standard Enterprise.

---

## Sources

- Anthropic: [Introducing Claude Fable 5 and Mythos 5](https://www.anthropic.com/news/claude-fable-5-mythos-5)
- Anthropic product page: [Claude Fable 5](https://www.anthropic.com/claude/fable)
- Anthropic official statement (June 12): [Statement on the US government directive to suspend access to Fable 5 and Mythos 5](https://www.anthropic.com/news/fable-mythos-access)
- TechCrunch (June 12): [Anthropic's safety warnings may have just backfired — the government has pulled the plug on its most powerful AI](https://techcrunch.com/2026/06/12/anthropics-safety-warnings-may-have-just-backfired-the-government-has-pulled-the-plug-on-its-most-powerful-ai/)
- NBC News (June 12): [Anthropic suspends new AI models after government directive](https://www.nbcnews.com/tech/tech-news/anthropic-suspends-new-ai-models-fable-mythos-government-directive-rcna349901)
- Axios (June 12): [Trump admin blocks foreign access to Anthropic's most powerful AI models](https://www.axios.com/2026/06/12/anthropic-trump-mythos-fable-national-security)
- The Decoder: [Anthropic releases Claude Fable 5 and Mythos 5 with major gains in coding and science](https://the-decoder.com/anthropic-releases-claude-fable-5-and-mythos-5-with-major-gains-in-coding-and-science/)
- TechCrunch: [Anthropic's Fable 5 can make weirdly fun video games with the click of a button](https://techcrunch.com/2026/06/09/anthropics-fable-5-can-make-weirdly-fun-video-games-with-the-click-of-a-button/)
- Fortune (June 10): [Anthropic accused of 'secret sabotage' as Claude Fable 5 silently limits capabilities for AI researchers and developers](https://fortune.com/2026/06/10/anthropic-accu-claude-fable-5-limits-capabilities-ai-researchers-developers/)
- Fortune (June 11): [After backlash, Anthropic says its AI will now tell users when their request is being rejected or downgraded for national security concerns](https://fortune.com/2026/06/11/anthropic-fable-5-silent-downgrade-backlash-national-security-transparency/)
- Decrypt: [The Internet Is Furious at Anthropic After Claude Fable 5 Release](https://decrypt.co/370688/internet-furious-anthropic-claude-mythos-fable-5)
- Latent Space: [AINews — Anthropic Claude Fable 5: Mythos but Safe, with Controversial Terms](https://www.latent.space/p/ainews-anthropic-claude-fable-5-mythos)
- Decrypt: [Anthropic Apologizes for Claude Fable 5 Secret Censorship — But the Fix Has a Catch](https://decrypt.co/370831/anthropic-apologizes-claude-fable-5-secret-censorship)
- Gizmodo: [Anthropic Apologizes For One of the Guardrails on Its Fable 5 Model, and Will Change It](https://gizmodo.com/anthropic-apologizes-for-one-of-the-guardrails-on-its-fable-5-model-and-will-change-it-2000770365)
- Simon Willison: [Anthropic Walks Back Policy That Could Have 'Sabotaged' AI Researchers Using Claude](https://simonwillison.net/2026/Jun/11/anthropic-walks-back-policy/)
- TechCrunch (June 13): [Amazon CEO reportedly raised Anthropic model concerns before government crackdown](https://techcrunch.com/2026/06/13/amazon-ceo-reportedly-raised-anthropic-model-concerns-before-government-crackdown/)
- Fortune (June 13): [Research leading to US export restrictions on top Anthropic models was for defense, cybersecurity CEO says](https://fortune.com/2026/06/13/anthropic-fable-mythos-models-commerce-deparment-export-restrictions-jailbreak-defense-prompting/)
- Fortune (June 15): ['Fix this code.' The three words that led the US government to ban Anthropic's Fable and Mythos](https://fortune.com/2026/06/15/fix-this-code-three-words-behind-us-government-shut-down-anthropic-fable-mythos-ai-models-katie-moussouris-open-letter/)
- Bloomberg (June 15): [Anthropic Set to Meet Trump Officials Over AI Security Concerns](https://www.bloomberg.com/news/articles/2026-06-15/anthropic-set-to-meet-trump-officials-over-ai-security-concerns)
- Axios (June 15): ['They screwed us': Personality clashes sent Anthropic's models offline](https://www.axios.com/2026/06/15/anthropic-white-house-fable-mythos)
- The Verge (June 15): [All the news about Anthropic's new AI fight with the White House](https://www.theverge.com/ai-artificial-intelligence/950026/anthropic-fable-mythos-ban-ai-shutdown)
- Bloomberg (June 15): [Cohere Seeing Inbounds After Washington's Anthropic Scrutiny](https://www.bloomberg.com/news/videos/2026-06-15/cohere-seeing-inbounds-after-dc-s-anthropic-scrutiny-video)
- Bloomberg (June 16): [Lutnick's Letter to Anthropic Warned of Curbs on Top AI Models](https://www.bloomberg.com/news/articles/2026-06-16/lutnick-s-letter-to-anthropic-warned-of-curbs-on-top-ai-models)
- Bloomberg (June 16): [Read the Lutnick Letter That Led Anthropic to Disable Mythos](https://www.bloomberg.com/news/articles/2026-06-16/read-the-lutnick-letter-that-led-anthropic-to-disable-mythos)
- PYMNTS (June 16): [Cybersecurity Experts Ask Feds to Lift Restrictions on Mythos](https://www.pymnts.com/news/artificial-intelligence/2026/cybersecurity-experts-ask-feds-to-lift-restrictions-on-mythos/)
- Bloomberg (June 19): [Trump Says He Doesn't View Anthropic as US Security Threat After Commerce Order](https://www.bloomberg.com/news/articles/2026-06-19/trump-tells-axios-he-doesn-t-see-anthropic-as-us-security-threat)
- Axios (June 19): [Exclusive: Trump tells "The Axios Show" that Anthropic was a national security threat](https://www.axios.com/2026/06/19/trump-anthropic-national-security-the-axios-show)
- U.S. News & World Report (June 17): [Trump Says Negotiations With Anthropic Are 'Going Fine'](https://www.usnews.com/news/world/articles/2026-06-17/trump-says-negotiations-with-anthropic-are-going-fine)
- The Register (June 15): [Anthropic reserves right to check ID for Claude subs](https://www.theregister.com/ai-and-ml/2026/06/15/anthropic-reserves-right-to-check-id-for-claude-subs/5255804)
- Computerworld: [Anthropic's new privacy policy offers US consumers a way around the Fable ban](https://www.computerworld.com/article/4185515/anthropics-new-privacy-policy-offers-us-consumers-a-way-around-fable-ban-2.html)
- TechCrunch (June 19): [The US banned Anthropic's Fable 5 release, but the numbers don't seem to care](https://techcrunch.com/podcast/the-us-banned-anthropics-fable-5-release-but-the-numbers-dont-seem-to-care/)
- TechCrunch (June 21): [When the Trump administration cracks down on Anthropic, who benefits?](https://techcrunch.com/2026/06/21/when-the-trump-administration-cracks-down-on-anthropic-who-benefits/)
- Live status: [Is Fable Back? — Live status of Claude Fable 5 & Mythos 5 access](https://isfableback.org/)
- Fortune (June 27): [Anthropic's Mythos 5 AI model cleared by U.S. for wider use](https://fortune.com/2026/06/27/anthropic-mythos-5-ai-model-us-commerce-department-clearance-fable/)
- CNBC (June 26): [Trump admin allows Anthropic to release Mythos AI model to some companies, government agencies](https://www.cnbc.com/2026/06/26/us-government-anthropic-claude-mythos5-ai.html)
- Axios (June 27): [Scoop: Powerful Anthropic model, Fable 5, on track to return soon](https://www.axios.com/2026/06/27/anthropic-fable-5-return-soon)
- Bloomberg (June 26): [US Allows Trusted Partners to Use Anthropic's Mythos 5 AI Model](https://www.bloomberg.com/news/articles/2026-06-26/us-allows-trusted-partners-to-use-anthropic-s-mythos-5-ai-model)
- Anthropic newsroom (June 30): [Redeploying Fable 5](https://www.anthropic.com/news/redeploying-fable-5)
- Fortune (July 1): [Anthropic's AI models are back online after a two-week government standoff—settling the company and administration into a fragile truce](https://fortune.com/2026/07/01/anthropic-fable-mythos-ai-models-restored-trump-administration-export-controls/)
- The Decoder (July 18, 2026): [Anthropic slashes Claude Fable 5 limits in Max and Team Premium and pushes Pro users toward API pricing](https://the-decoder.com/anthropic-slashes-claude-fable-5-limits-in-max-and-team-premium-and-pushes-pro-users-toward-api-pricing/)
- AI Weekly (July 18, 2026): [Anthropic caps Claude Fable 5 in Max, cuts Pro users off](https://aiweekly.co/alerts/anthropic-caps-claude-fable-5-in-max-cuts-pro-users-off)
- TechTimes (July 18, 2026): [Claude Fable 5 Ends Subscription Limbo: Permanent for Max, Credits-Only for Pro](https://www.techtimes.com/articles/320905/20260718/claude-fable-5-ends-subscription-limbo-permanent-max-credits-only-pro.htm)
- Android Authority (July 1): [Anthropic's Fable 5 is back, and that's no fairy tale](https://www.androidauthority.com/fable-5-back-3683475/)
## Frequently asked questions

### Is Claude Fable 5 available right now?

Yes, but billing splits permanently on July 20, 2026. Through July 19, Fable 5 is available on Pro, Max, Team, and select Enterprise plans via credits. Starting July 20: Max and Team Premium subscribers keep Fable 5 included at 50% of their plans' weekly limits (which themselves are cut by ~33% when the bonus usage phase ends, so effective Fable 5 access is roughly a third of the original launch allowance). Pro and Team Standard subscribers lose subscription access to Fable 5 entirely — they receive a one-time $100 usage credit for API access ($10/M input, $50/M output), then pay API rates ongoing. Mythos 5 remains available to 100+ trusted US companies and federal agencies through Project Glasswing.

### What changes for my Claude plan on July 20, 2026?

Anthropic made the Fable 5 billing split permanent on July 20. If you're on Max or Team Premium: Fable 5 stays included, but the bonus usage phase ends the same day — reducing your base limits by ~33%, then Fable 5 draws from 50% of those reduced limits. You still have the model, just less of it. If you're on Pro or Team Standard: you lose subscription access to Fable 5. You'll get a one-time $100 usage credit for Fable 5 via API ($10 per million input tokens, $50 per million output tokens — double the Opus 4.8 API price), then pay API rates from there. Sonnet 5 and Opus 4.8 remain fully available on all plans and are unaffected by these changes.

### Why did the US government shut down Claude Fable 5?

The US government issued an export control directive on June 12, 2026 citing national security concerns and a claimed method of jailbreaking Fable 5. Anthropic disputed this characterization. June 13 update: the Wall Street Journal reported that the research was conducted by Amazon researchers and briefed to government officials by Amazon CEO Andy Jassy, who said the work surfaced information 'that could be used in cyberattacks.' Security expert Katie Moussouris (Luta Security) reviewed the research and described it as 'Defense Oriented Prompting (DOP)' — a defensive security technique, not an offensive jailbreak. Anthropic complied with the directive but published a statement disagreeing with the scope and process.

### What did people use instead of Fable 5 during the suspension?

Claude Opus 4.8 and Sonnet 4.6 remain fully available and unaffected by the directive. Opus 4.8 was Anthropic's top publicly available model before Fable 5 launched and is capable for complex professional work: long documents, multi-step workflows, code, and detailed analysis. For everyday tasks — drafting, research, summaries, client communications — Sonnet 4.6 is fast and reliable. API users should swap any claude-fable-5 model strings for claude-opus-4-8.

### What specifically triggered the government's jailbreak concern?

Fortune's June 15 reporting reveals the specific technique: Amazon researchers found that Fable 5 refused 'review this code for security issues' but complied with 'fix this code' — producing software patches that could theoretically expose vulnerability information. The model's guardrails were tuned to block security-analysis framing but not code-repair framing. Security expert Katie Moussouris, who reviewed the Amazon research, describes this as 'Defense Oriented Prompting' — a defensive technique — rather than an offensive exploit.

### Will Claude Fable 5 come back?

It already has. Fable 5 was restored globally on July 1, 2026, following a June 30 lifting of export controls — 19 days after the suspension. Anthropic committed to several obligations to close the standoff: expanded early access to frontier models for federal security review, rapid jailbreak information sharing with government partners, dedicated joint research teams, and contributing to a shared voluntary jailbreak severity scoring framework with Amazon, Microsoft, and Google.

### What is Claude Fable 5, and how does it compare to Claude Opus 4.8?

Claude Fable 5 is Anthropic's newest top model, released June 9, 2026. It represents a new generation — Mythos-class — that sits above the Opus tier. On coding benchmarks, it scores 80.3% on SWE-Bench Pro versus 69.2% for Opus 4.8, and 29.3% on FrontierCode versus 13.4% for Opus 4.8. It is meaningfully better at long multi-step tasks, document-heavy reasoning, and vision analysis. Access was suspended June 12–July 1 under a US government export control directive; it is now available globally.

### What is Claude Mythos 5, and can I use it?

Claude Mythos 5 is a more capable but restricted version of the same generation as Fable 5. As of July 1, 2026, Mythos 5 is available to 100+ trusted US companies and federal agencies through Anthropic's Project Glasswing program — cybersecurity companies and critical infrastructure providers including Apple, Google, Cisco, Nvidia, and Microsoft. Non-US nationals at approved organizations are also eligible. For most professionals using Claude.ai or the API, Mythos 5 is not available; Fable 5 (now restored globally) is the model for you.

### Does Claude Fable 5 change my data privacy settings?

Yes — Anthropic requires 30-day data retention for Fable 5 safety monitoring, a change from zero-retention policies some API users previously maintained. This requirement is now active (Fable 5 restored July 1). Enterprise teams with strict data-handling requirements should review Anthropic's updated data terms before deploying Fable 5.

### Does Claude Fable 5 ever give deliberately weakened responses without telling me?

Not anymore — Anthropic reversed this on June 11, 2026, before the June 12 suspension. Flagged requests now visibly fall back to Opus 4.8 on Claude.ai rather than returning a silently degraded response. API calls return an explicit reason for refusal. The underlying restriction — preventing use of Claude to build competing AI systems or assist foreign adversaries — remains in place, but triggers visibly.

### When Fable 5 came back, was it restricted like OpenAI's GPT-5.6 Sol?

No — Fable 5 returned to all users globally on July 1, 2026, not just a narrow partner tier. GPT-5.6 Sol spent its first 13 days (June 26–July 9, 2026) restricted to roughly 20 government-approved partners; Anthropic took the broader global path from day one of restoration. That said, subscription access has been narrowed since: through a series of extensions (July 7, July 12, July 19), billing eventually settled into a permanent split effective July 20 — Max and Team Premium keep Fable 5 at reduced limits; Pro and Team Standard lose subscription access and move to API credits. The 30-day data retention requirement for safety monitoring remains in effect for API users.

---

*Canonical version: https://theaicareerlab.com/blog/claude-fable-5-for-professionals*
*This document is the Markdown companion served for AI crawlers and answer engines. See the canonical URL for the rendered version with navigation, related content, and interactive elements.*