Anthropic confirms – Anthropic has confirmed it will bring Mythos-class Claude models to the general public in the coming weeks, after delaying the rollout earlier this year over security risks to public and private software. Mythos was first announced in April as a restricted mod
When Anthropic first unveiled Mythos in April, it wasn’t the kind of announcement that made everyone reach for the button. The company framed it as a restricted model—and then held it back from wider release, pointing to security risks to both public and private software.
That restraint now looks temporary. Anthropic has confirmed it plans to bring Mythos-class models to the general public after making changes intended to prevent misuse. In a blog post. the company said it is “making swift progress on developing these safeguards” and expects to be able to bring Mythos-class models “to all our customers in the coming weeks. ” without committing to an exact date.
Mythos itself was announced in April as a restricted model, made available only to select companies, including security researchers. At the time, Anthropic explicitly warned that the power of the model could cut in different directions depending on who gets access first.
“The advantage will belong to the side that can get the most out of these tools,” Anthropic warned in April when it announced the Mythos model.
“In the short term, this could be attackers, if frontier labs aren’t careful about how they release these models. In the long term, we expect it will be defenders who will more efficiently direct resources and use these models to fix bugs before new code ever ships.”
For months. that tension has been familiar across the industry: powerful AI tools often stay behind controlled doors until companies believe they have guardrails strong enough to limit the damage that misuse can cause. Anthropic’s decision to delay the public rollout was part of that logic. with the company citing “security” risks tied to the Mythos model and choosing not to move forward publicly.
The new confirmation suggests Anthropic believes it has closed that gap. The company says Mythos-class models bring major improvements in code reasoning and autonomy—performance described as far above Claude’s current flagship model, Opus 4.8.
Still. the rollout comes with unanswered questions about how identical the public version will be to what has already been tested behind the scenes. Anthropic says it is already allowing a small number of organizations to use the “Claude Mythos preview” for cybersecurity work. but it remains unclear whether the exact same model will be released to the public.
There’s also a trace of what might have been a wider preview than the company intended. The “Mythos-preview” model briefly appeared for some users on Claude Code before it was taken offline.
Putting those details together. the story reads like a moving target rather than a fixed launch date: Mythos started as a restricted release for select companies in April. Anthropic paused broader access after warning about security risks. and now the company is signaling that safeguards are in place enough to expand access “in the coming weeks. ” with the specifics of what users will get still partly unsettled.
Anthropic Claude Mythos-class models AI safety cybersecurity guardrails Opus 4.8 Claude Code software security