Anthropic Says a Mythos-Class AI Model Will Be Available Soon

Anthropic has released Claude Opus 4.8, a new version of its large language model described as a 'modest but tangible improvement' over the previous Opus 4.7. The company also announced it is making progress on releasing a public version of its Claude Mythos Preview model, which was previously only available as a preview to select users.

According to the report, the Opus 4.8 update focuses on incremental refinements rather than major architectural changes. Anthropic's statement emphasized that the new model offers concrete enhancements in performance and reliability compared to its predecessor, though the company did not provide specific metrics or benchmarks. The Mythos model, which has been in preview, represents a more advanced system that the company is working to bring to the public.

The release of Opus 4.8 and the ongoing development of Mythos illustrate Anthropic's accelerated release cadence as it competes with other frontier labs. The company, which has built its reputation on safety and alignment research, continues to iterate on its models while hinting at more advanced future systems. Anthropic has not provided a specific timeline for the public release of Mythos, stating only that progress is being made toward that goal.