Claude Fable 5 is Anthropic’s most capable publicly available AI model, released June 9, 2026 — the first model in its new Mythos class, a tier above Claude Opus. It costs $10 per million input tokens and $50 per million output, and is included on paid Claude plans through June 22.
Anthropic announced the launch on Tuesday. Until now, Mythos-level capability was locked behind Project Glasswing, a restricted cyberdefense program run with the US government. This is the first time anyone can touch it.
Claude Fable 5 Comes With a Built-In Leash
Fable 5 ships with safety classifiers covering cybersecurity, biology and chemistry, and distillation attempts. When triggered, the response is handed to Claude Opus 4.8 instead, and users are notified. Opus is no consolation prize — it was Anthropic’s flagship until Tuesday.
Anthropic says the fallback fires in under 5% of sessions, and admits the classifiers are tuned conservatively enough to catch some harmless requests. An external bug bounty ran over 1,000 hours of red-teaming without producing a universal jailbreak.
Stripe’s Two-Month Migration Took One Day
In early testing, Stripe reported Fable 5 completed a codebase-wide migration across 50 million lines of Ruby in one day — work estimated at over two months for a full engineering team.
Cursor, GitHub, Cognition, and Replit all reported state-of-the-art results on their internal coding benchmarks, per TechCrunch’s launch coverage. Anthropic also says Fable 5 beat Pokémon FireRed using raw screenshots alone, with no helper harness — something no previous Claude model managed.
The June 23 Catch
Fable 5 is included on Pro, Max, Team, and seat-based Enterprise plans only through June 22. From June 23, it will require usage credits until Anthropic adds capacity.
API access is unrestricted from day one via claude-fable-5, at less than half the price of Mythos Preview.
Mythos 5 Stays Behind Glass
The same model also shipped as Claude Mythos 5, with cyber safeguards lifted — but only for vetted Glasswing partners. Anthropic calls it the strongest cybersecurity model in the world, and says a trusted access program for security and biology researchers is next.
There’s a privacy trade-off too: all Mythos-class traffic now carries mandatory 30-day data retention, even for business customers — a first for Anthropic, justified as a defense against novel jailbreaks.
Frequently Asked Questions
What is Claude Fable 5?
It’s Anthropic’s most powerful publicly available AI model, launched June 9, 2026 — the first Mythos-class model released for general use, with added safety classifiers.
How much does Claude Fable 5 cost?
$10 per million input tokens and $50 per million output tokens on the Claude API. It’s included on paid Claude plans at no extra cost through June 22, 2026.
What is the difference between Claude Fable 5 and Mythos 5?
Same underlying model. Fable 5 carries safety safeguards for general release; Mythos 5 has cyber safeguards lifted and is restricted to vetted Project Glasswing partners.
Why does Claude Fable 5 fall back to Opus 4.8?
Classifiers route requests involving cybersecurity, biology and chemistry, or distillation attempts to Claude Opus 4.8 instead — in under 5% of sessions, per Anthropic.
Is Claude Fable 5 available now?
Yes — live on the Claude API, Amazon Bedrock, and Microsoft Foundry, and on paid Claude plans through June 22 before moving to usage credits.
The launch lands days after Anthropic’s Mythos hacking-capability disclosure and mid-IPO — see Anthropic’s IPO filing and $965B valuation. Follow WithO2 for what changes after June 23.
