Anthropic released Claude Fable 5, a “Mythos-class” model designed to be more capable than any generally available model it has made, but with major safety routing built in. When Fable 5 detects requests tied to cybersecurity, biology and chemistry, or distill
For a company that spent weeks inviting the market to wonder how far its “Mythos”-level systems could go, Anthropic’s latest release arrives with a catch you can feel in the wording.
On Tuesday, Anthropic said it was releasing a “Mythos-class” model called Claude Fable 5—but the safeguards attached to that capability were significant enough that the company effectively treated the changes as part of what made it safe to use.
Anthropic said its safeguards are designed to prevent misuse. “Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage,” Anthropic said in its announcement.
So the system doesn’t just answer. It reroutes. Anthropic said that some queries on certain topics will instead be ported to its Opus 4.8 model. The safeguards are initially tuned to be more conservative. and they trigger when Fable 5 detects requests concerning cybersecurity. biology and chemistry. or distillation. If that happens, Anthropic said users will be informed that their request is being routed through Opus 4.8.
The release also keeps Anthropic’s earlier trail alive. The company is continuing Project Glasswing — first announced in April — which gave early access to the first version of Mythos. Claude Mythos Preview. to a limited number of tech. cybersecurity. and financial companies. Anthropic now said a “small group of cyberdefenders and infrastructure providers” will get access to Claude Mythos 5. which it described as the same model as Fable 5 “but with the safeguards lifted in some areas.”.
Anthropic said Mythos 5 “has the strongest cybersecurity capabilities of any model in the world.” It added: “Soon, we intend to expand access to Mythos 5 through a broader trusted access program.”
Even with those constraints on Fable 5, Anthropic says the baseline remains high. The company said Fable 5 “exceed those of any model we’ve ever made generally available.”
In performance terms, Anthropic framed Fable 5 as a long-task champion. The company said it is “state-of-the-art on nearly all tested benchmarks of AI capability. ” listing “exceptional performance in software engineering. knowledge work. vision. scientific research. and many other areas.” It added: “The longer and more complex the task. the larger Fable 5’s lead over our other models.”.
For two specific application categories, Anthropic pointed to external evaluations. It said Stripe found that Fable 5 compressed “months of engineering into days.” For financial tasks. Anthropic said IMC found that Fable 5 “aced their trading-analysis evaluations nearly across the board. ” including “factual lookup. conceptual reasoning. root-cause analysis. and expected-value analysis.”.
The model naming follows Anthropic’s existing approach of treating AI systems as forms of composition. Fable is from the Latin “fabula,” meaning “that which is told,” akin to the Greek “mythos,” Anthropic said. It also said Fable was chosen because it shares similarities with Mythos.
This timing matters beyond product design. Anthropic’s release comes at a critical moment for the company as it moves toward public markets. The company recently announced that it had confidentially filed a draft S-1, officially kicking off the march toward a blockbuster IPO.
The competitive landscape is moving at the same pace. OpenAI, Anthropic’s chief rival, announced on Monday that it too had confidentially filed its S-1. SpaceX, which includes Elon Musk’s xAI, is set to go public on Friday when it begins trading on the Nasdaq.
And the scrutiny around frontier AI has intensified. Anthropic’s initial announcement of Mythos spooked financial markets and governments worldwide, raising concerns that AI models had advanced to the point where they could now uncover major vulnerabilities in software and cybersecurity.
That concern has also spilled into Washington’s policy moves. President Donald Trump recently signed an executive order that allows AI companies to voluntarily give the federal government access to frontier AI models up to 30 days before their release. The order explicitly prohibits the government from imposing a mandatory review on AI companies.
So with Claude Fable 5, Anthropic is essentially asking for trust on two different levels at once: broad public availability with routing safeguards for high-risk requests, and separate pathways for tightly managed access through Project Glasswing and a future “trusted access program” for Mythos 5.
What’s changed is not the ambition. It’s the wiring.
Anthropic Claude Fable 5 Mythos-class model Opus 4.8 Project Glasswing Claude Mythos 5 IPO draft S-1 Stripe IMC AI safeguards cybersecurity model