Anthropic's most capable generally-available model, Claude Fable 5, went dark for close to three weeks after a US government directive suspended access — and came back online globally on July 1, 2026. Its sibling model, Claude Mythos 5, is a different story entirely: it was never meant for the general public in the first place. Here is what actually happened, in order.
The Timeline
June 9, 2026 — Launch. Anthropic released Claude Fable 5 as its next-generation model for "days-long, complex, and asynchronous tasks previous models couldn't sustain," generally available on the Claude API, Claude Platform on AWS, Amazon Bedrock, Google Cloud, and Microsoft Foundry. Alongside it, Anthropic introduced Claude Mythos 5 — a model sharing Fable 5's specs and pricing, but released only to approved partners inside Project Glasswing, an invitation-only cybersecurity initiative, not to the general public.
June 12, 2026 — Suspension. Anthropic issued a public statement on a US government directive to suspend access to both Fable 5 and Mythos 5. Anthropic's own account of the disruption was direct: "We apologize for this disruption to our customers and are working to restore access as soon as possible."
June 30, 2026 — Redeployment plan announced. Anthropic published "Redeploying Fable 5," committing to further-enhanced safeguards and proposing an industry-wide framework for scoring jailbreak severity, developed together with Amazon, Microsoft, Google, and other Project Glasswing partners. The post confirmed Fable 5 would return globally on July 1.
July 1, 2026 — Fable 5 restored. Access to Claude Fable 5 was restored globally, as announced. Anthropic's own update was simple: "Access to Claude Fable 5 has been restored. It brings 5th-generation intelligence to your most ambitious coding and professional work."
What Claude Fable 5 Actually Is
Fable 5 is positioned as Anthropic's most capable generally available model — built for ambitious, long-running, largely unsupervised work rather than everyday quick-turnaround tasks. In an agent harness (like Claude Code or a managed agent runtime), it can plan across stages, delegate to sub-agents, write its own tests, and check its work using vision — running for days at a time on a single complex objective such as a large codebase migration or a multi-stage research task. It carries a 1M-token context window and is priced at $10 per million input tokens and $50 per million output tokens, with the standard 90% prompt-caching discount available.
Because a model this capable raises real risk in domains like cybersecurity and biology, Fable 5 ships with dedicated safeguards: queries flagged in these areas are automatically rerouted to Claude Opus 4.8 instead of being answered directly, and — importantly for anyone budgeting API usage — you are not charged Fable-tier pricing for rerouted requests.
Why Claude Mythos 5 Is a Different Situation Entirely
Mythos 5 was never intended for public or developer access, and that has not changed. It exists inside Project Glasswing, a defensive-cybersecurity initiative Anthropic launched in April 2026 together with AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. The premise: Anthropic's research found that frontier models had crossed a threshold where they could find and exploit software vulnerabilities at a level that rivals top human security researchers — the preview model autonomously found a 27-year-old vulnerability in OpenBSD and a 16-year-old bug in FFmpeg, among others. Rather than release that capability broadly, Anthropic restricted it to vetted partners hardening critical infrastructure and open-source software, at a separate access tier priced at $25/$125 per million input/output tokens for participants.
If you are a DIGIT client or developer wondering whether you can request Mythos 5 access for a project: the honest answer is no, not through a standard commercial relationship. It is available only through Anthropic's Project Glasswing partner program, aimed specifically at organizations doing large-scale defensive security and open-source infrastructure work — sign-up for updates on trusted-access programs is handled directly through Anthropic.
What This Means Practically
For teams building products, the relevant news is simply that Fable 5 is back and available exactly as it was before the suspension — no client-facing changes to the API surface, pricing, or integration pattern. For anything security-research related that might have prompted interest in Mythos 5 specifically, that path runs through Anthropic's Glasswing program directly, not through general API access.
If you are scoping a long-running agentic project (a large migration, a multi-day research task, or an ambitious coding project) and want to evaluate whether Fable 5 is the right model tier versus Sonnet 5 or Opus 4.8, reach out at info@digit.com.pk.