Claude Fable 5 Is Now Public: Inside Anthropic's Most Powerful Model Yet
TL;DR
Anthropic just made its Mythos-class model publicly available for the first time. Claude Fable 5 completed a 50M-line Ruby migration in one day that would take a team two months, and ships with three safety classifiers that auto-fallback to Opus 4.8.
Claude Fable 5 launched on June 9, 2026, the first time Anthropic has made a Mythos-class model available to the general public.
For months, the Mythos series stayed behind closed doors, restricted to select government and enterprise partners. The same underlying model now ships to the API and subscription plans, wrapped in three safety classifiers.
One Day Instead of Two Months
Stripe put Fable 5 to work on a 50-million-line Ruby codebase migration during testing. The estimated timeline for an engineering team: two months. Fable 5 finished in one day.
A 60x compression ratio on a real-world production task changes how teams think about headcount and project scope, not just how they think about tooling. Genspark’s evaluation found Fable 5 beat every other model on UI design and game coding tasks. Hex Analytics reported it was the first model to cross 90% on their core analytics benchmark.
Vision capability came with an unconventional test: using only raw screenshots with no helper tools, Fable 5 completed Pokémon FireRed from start to finish. On Cognition’s FrontierCode evaluation, it ranked highest among all frontier models tested.
For scientific research, Anthropic’s internal molecular biology experts ran blind comparisons and preferred Fable 5’s outputs 80% of the time, describing it as “the first model to consistently produce genuinely novel molecular biology hypotheses.” Drug design workflows accelerated roughly tenfold.
Three Classifiers That Define What It Will and Won’t Do
Fable 5 ships with three safety classifiers. When any one triggers, the request automatically falls back to Claude Opus 4.8:
The cybersecurity classifier blocks offensive penetration testing and vulnerability exploitation requests. External red-teamers ran more than 1,000 hours of testing and found no universal jailbreak. Across 30 publicly known jailbreak techniques, Fable 5 maintained 100% compliance on single-turn harmful requests.
The biology and chemistry classifier is strict by design. Fable 5’s ability to predict adeno-associated virus properties outperforms specialized protein models. It is valuable for gene therapy research, but the dual-use risk led Anthropic to keep this capability gated until a trusted-access program launches.
The distillation prevention classifier blocks systematic capability extraction attempts aimed at training competing models.
The classifiers are tuned conservatively: on average, fewer than 5% of sessions trigger a fallback. Occasionally that means catching harmless requests. Anthropic’s official launch post frames this as the cost of careful rollout for a Mythos-class system.
Mythos 5 Runs on a Separate Track
Fable 5 and Mythos 5 share the same underlying model. The difference is which classifiers are active.
Mythos 5 is currently restricted to Project Glasswing partners, with the cybersecurity classifier removed. Glasswing launched in April and expanded in early June to roughly 150 organizations across 15+ countries, including US government agencies and authorized critical infrastructure providers.
A biology trusted-access program is also in planning, which would let academic and pharmaceutical researchers use Mythos 5 with the biology classifier lifted while keeping cybersecurity guardrails intact.
Pricing and Rollout Timeline
Fable 5 costs $10 per million input tokens and $50 per million output tokens, roughly half the price of Mythos Preview but double Opus 4.8.
Subscription plan access (Pro, Max, Team, and seat-based Enterprise) is free from June 9 through June 22. Starting June 23, it requires usage credits purchased separately. API and consumption-based enterprise access is immediate. The model is also available on Amazon Bedrock from day one.
Anthropic has implemented a 30-day data retention policy for Mythos-class traffic, used only for safety review and not for model training.
The timing matters: Anthropic confidentially filed its S-1 with the SEC on June 1. Making Fable 5 publicly available two weeks before the subscription cutoff, with named enterprise adopters like Stripe, Rakuten, and Genspark, gives prospective investors a concrete view of commercial traction before any public filing.
If this was useful, subscribe to the newsletter for weekly AI PM insights and GenAI case studies.
Further reading:
Related Articles
White House Demands Zero Jailbreaks for Fable 5: Security Experts Say It's Impossible
Day 7 of the Fable 5 ban: the White House demands the model be completely jailbreak-proof before it relaunches. Security experts are unanimous: that's technically impossible for any frontier LLM, and Dario Amodei has already refused both of the government's proposed fixes.
US Orders Anthropic to Pull Fable 5 and Mythos 5: A Narrow Jailbreak That Took Down Its Most Powerful Models
The US Commerce Department ordered Anthropic to suspend its two most capable models, Fable 5 and Mythos 5, citing a narrow jailbreak tied to cybersecurity capabilities. Anthropic complied. Then it pushed back.