Claude Mythos 5:
What It Is and Who Actually Gets Access
In one line: Mythos 5 is Fable 5 with the guardrails lifted — the same underlying model, with restrictions removed in some areas, billed as having the strongest cybersecurity capabilities of any model in the world. And it is not publicly available.
From Mythos Preview to Mythos 5: the timeline
April 2026: Anthropic previewed Mythos (Claude Mythos Preview), its strongest internal model — and kept it limited to a handful of partners because it was considered too skilled at finding and exploiting software vulnerabilities.
June 9, 2026: Anthropic shipped that single frontier as two products:
- Fable 5 (
claude-fable-5) — the public release: same training, frontier capabilities, plus safety classifiers that hand cybersecurity, biology, chemistry, and distillation queries to Opus 4.8. - Mythos 5 — the same underlying model with safeguards lifted in some areas. It upgrades Claude Mythos Preview and is deployed through Project Glasswing, in collaboration with the US government, to a small group of vetted cyberdefenders and infrastructure providers.
This split — one frontier, two products — is Anthropic's compromise between "too dangerous to release" and "the public deserves frontier capability."
Mythos 5 vs Fable 5: what's actually different
| Fable 5 | Mythos 5 | |
|---|---|---|
| Base model | Identical (same training, same capability baseline) | |
| Safeguards | Full (sensitive queries fall back to Opus 4.8) | Lifted in some areas |
| Cybersecurity capability | Classifier-limited | Strongest of any model (per Anthropic) |
| Who can use it | Everyone (API, cloud, claude.ai) | Vetted cyberdefenders & infrastructure providers |
| How to get it | Sign up and call the API | Project Glasswing deployment; no public application |
| Pricing | Same: $10 / $50 per million input/output tokens | |
A common misconception worth killing: Mythos 5 is not "smarter" than Fable 5. They're the same model — the difference is only what the safeguards allow. For coding, writing, analysis, and agent work, Fable 5 already gives you the full Mythos-class capability.
How to apply for Mythos 5 (honestly: you almost certainly can't)
As of June 2026 there is no public application channel. Initial deployment goes to cyberdefense and critical-infrastructure organizations vetted under Project Glasswing, in collaboration with the US government. Individual developers and ordinary companies are not in scope.
The practical takeaway
For 99.9% of users, the only way to "use Mythos" is to use Fable 5 — it is the Mythos Anthropic built for the public. Unless your work trips the <5% sensitive-domain classifiers, the two are indistinguishable. Full Fable 5 guide here.
Use the public Mythos — Fable 5
OmniaKey: Fable 5 at $3/$15 (limited-time, ~70% off) · official-compatible endpoint · no card required
FAQ
What is Claude Mythos 5?
Anthropic's restricted frontier model, released June 9, 2026 alongside Fable 5: the same underlying model with safeguards lifted in some areas, available only to vetted cyberdefenders via Project Glasswing.
Is Mythos 5 more capable than Fable 5?
Same base model. Mythos 5 only does more in domains Fable 5's safeguards restrict (chiefly cybersecurity). Everywhere else they're equivalent.
Can I apply for Mythos 5 access?
No public channel exists. It's deployed to vetted cyberdefense and infrastructure organizations under Project Glasswing. The public counterpart is Fable 5 (claude-fable-5).
Will Mythos 5 ever be public?
No timeline announced. Anthropic's pattern is to ship realigned, safeguarded derivatives publicly (that's exactly what Fable 5 is) rather than release unguarded frontier models.