✳ the wire · analysis

Dario Amodei says Anthropic had Mythos-class models in February and held them back until June for cyber safeguards

Mythos 5confirmedby ArtificialWatch

An email the second Anthropic’s next model answers on a public API, plus every frontier launch · no card

ArtificialWatch title card · verified against a primary source

Anthropic CEO Dario Amodei says the company first had Mythos-class models in February 2026 but did not make them generally available until June, once stronger cyber safeguards were in place. Until then it offered them only to a limited set of partners, to focus on cyberdefense. He wrote it on X on October 21, in a long reply to Nathan Calvin, who had asked in an October 19 post whether Anthropic still acts like an AI safety organization or mainly like a company racing to build superhuman AI first.

WHAT HE SAID

Amodei calls the delay costly. Releasing to everyone would have been the economically rational move, he writes, and Anthropic likely gave up a significant part of a several-month lead. His argument is that companies used the time to secure their systems before cyber-capable models became widely available.

It is one of nine decisions he lists as costly to Anthropic and good for safety. Among the others: strict biological safeguards, after concerning results in bio-uplift trials and blocked attempts to misuse its models for bioweapons, at the price of false positives biologists complain about; restricting sales to PRC-controlled companies, including through third countries, which he says, as far as he knows, no other frontier company does; work against distillation, which he says is not a major commercial threat but raises national security and safety concerns; and putting 5% of the company, possibly more over time, into a public benefit fund, announced this week.

He also writes that Anthropic has more than 5,000 people, that AI-driven acceleration is nearing the point where it is hard for safety to keep up, which drives its thinking about "pacing", and that he will have more to say soon about how Anthropic will act on its values.

WHAT IT MEANS FOR THE MODELS

The dates match ArtificialWatch's record: Mythos 5 reached general release on June 9, alongside Fable 5. Its successor, Mythos 5.1, released on September 1, is still limited to vetted organizations through Anthropic's life-science and cyber verification programs.

WHAT WE DO NOT KNOW

What Amodei plans to announce, and whether "pacing" means slower releases. Anthropic has not announced Mythos 6 or given a date for it.

Source: Dario Amodei on X (post read directly) · Nathan Calvin on X · @kimmonismus on X ↗ · Mythos 5 tracker · the bench index

sweeping every 60 seconds

Hear the minute Anthropic’s next model drops — not the minute we write it up.

Claude Opus 5 went live at 16:51 UTC. The alert was in subscribers' inboxes at 16:52.

  • Free forever
  • No card
  • Unsubscribe in one click

← back to the wire