June 10, 2026 — Anthropic has launched Claude Fable 5, making its most capable class of AI models available to the general public for the first time — alongside a limited rollout of the even more powerful Claude Mythos 5 for a select group of trusted partners.
The release marks a significant milestone in the AI industry. Fable 5 belongs to the same family as Anthropic's Mythos-class models, which earlier this year drew widespread attention — and concern — for their superhuman ability to identify and exploit cybersecurity vulnerabilities. Now, the company is bringing that frontier capability to everyday users, albeit with meaningful restrictions in place.
What Is Fable 5?
Fable 5 is Anthropic's new flagship public model, outperforming all previously released Claude models across nearly every benchmark. Anthropic says it excels in software engineering, knowledge work, scientific reasoning, and vision-based tasks. The longer and more complex a task, the wider Fable 5's advantage over earlier models.
Early enterprise testing has been striking. Stripe reported that the model compressed months of engineering work into days, completing a codebase-wide migration across a 50-million-line Ruby codebase in roughly 24 hours — a task that would have taken an entire team of engineers over two months by hand.
In finance, Fable 5 topped Hebbia's senior-level reasoning benchmark, showing significant gains in document analysis, chart interpretation, and problem-solving. Trading firm IMC said the model aced their internal evaluations across factual lookup, root-cause analysis, and expected-value reasoning.
Vision capabilities have also taken a leap. Fable 5 can reconstruct a web application's source code from screenshots alone and — in a memorable demo — completed the classic Pokémon FireRed using nothing but raw game screenshots, with no maps, navigation aids, or additional scaffolding. Previous Claude models required a complex helper harness to accomplish the same feat.
The Safety Tradeoff
Releasing a model this capable doesn't come without risk. Anthropic has been candid that Fable 5's raw capabilities — particularly in cybersecurity — could be misused to cause serious harm. To address this, the company has built in hard safety limits: queries touching sensitive areas like cybersecurity, biology, chemistry, and distillation are automatically rerouted to Claude Opus 4.8, the company's next-most-capable model, rather than answered directly.
Anthropic acknowledges the safeguards are tuned conservatively and will sometimes catch benign requests — but says this occurs in fewer than 5% of sessions on average. The company says it is actively working to reduce false positives as quickly as it can, with more capable models expected in the coming months.
What About Mythos 5?
Running alongside the public launch is a quieter, more restricted one. Claude Mythos 5 — the same underlying model as Fable 5, but with some of the safety limits lifted — is being deployed through Project Glasswing, Anthropic's collaboration with the U.S. government, as an upgrade to the earlier Claude Mythos Preview.
Mythos 5 currently has access limited to a small group of cyber defenders and critical infrastructure providers. Anthropic says it holds the strongest cybersecurity capabilities of any AI model in the world. Through Project Glasswing, it has already been used to help defenders secure critically important software — the same class of vulnerabilities the Mythos family was initially flagged for being able to exploit.
In life sciences, Mythos 5 is showing early promise that goes beyond cybersecurity applications. Anthropic researchers used the model to accelerate aspects of the drug design process by roughly ten times. In one controlled study, Mythos 5 matched or outperformed skilled human operators on protein design tasks — selecting binding sites, running bioinformatics tools, and recovering from failures autonomously. Nine of the 14 protein targets tested yielded strong drug design candidates that Anthropic says it is actively investigating.
The model has also begun producing what Anthropic describes as novel, compelling scientific hypotheses in molecular biology — with scientists preferring Mythos's output over Opus-class models roughly 80% of the time in blinded comparisons. One Mythos-generated hypothesis about an E. coli protein mechanism was independently corroborated by a separate academic lab.
Anthropic says it intends to expand Mythos 5 access through a broader trusted-access program in the months ahead.
Pricing and Availability
Both models are priced at $10 per million input tokens and $50 per million output tokens — less than half the cost of Claude Mythos Preview. Fable 5 is available today through the Claude API, Amazon Bedrock, Microsoft Foundry, and GitHub Copilot. Mythos 5 access remains restricted to Project Glasswing participants for now.
The names carry weight: fable traces to the Latin fabula, meaning "that which is told," while mythos is its Greek counterpart. Anthropic says the different names reflect a meaningful distinction — not just in safeguards, but in the responsibility that comes with each model's reach.
