Anthropic confirms Claude Mythos-class models will roll out to the public
Update internal AI strategy to incorporate Mythos when released and monitor Anthropic for release details.
Update internal AI strategy to incorporate Mythos when released and monitor Anthropic for release details.
Summary
Anthropic has announced that it will bring its Mythos‑class models to the general public after a delay caused by security concerns.
The Mythos model, first revealed in April as a restricted model, is considered far more powerful than the current flagship Opus 4.8, offering significant gains in code reasoning and autonomous task execution. Anthropic said it has built robust guardrails that mitigate misuse, allowing it to release the model to a limited preview set of organizations for cybersecurity work. The company will make Mythos available to all customers in the coming weeks, though no exact date has been set. A preview version briefly appeared on Claude Code before being taken offline. The announcement follows Anthropic’s earlier warning that attackers could exploit the model if guardrails were insufficient. The rollout will enable developers to leverage advanced AI capabilities while maintaining safety controls.
Key changes
- Mythos‑class models are far more powerful than Opus 4.8
- Provide advanced code reasoning and autonomous task execution
- Anthropic built robust guardrails to mitigate misuse
- Limited preview available to select organizations for cybersecurity work
- Public rollout planned for coming weeks, no exact date yet
- Preview briefly appeared on Claude Code before removal
- Anthropic warns attackers could exploit model if guardrails fail
- Release will enable developers to use advanced AI while maintaining safety