Artificial intelligence company Anthropic has launched two versions of its new AI model, making Claude Fable 5 available to the general public while restricting access to Claude Mythos 5 to selected partners.
The company announced the models on June 9, saying Fable 5 includes additional safeguards while Mythos 5, which is based on the same underlying architecture, will initially be available to organisations participating in its Project Glasswing programme.
According to Anthropic, Fable 5 is designed to handle most user requests, but queries involving areas such as cybersecurity, biology and chemistry are redirected to an older model, Claude Opus 4.8. The company said the approach was intended to provide access to higher levels of AI capability while limiting risks associated with misuse.
“We wanted to be able to provide this level of intelligence for general users in a safe manner,” said Dianne Penn, Anthropic’s head of product management, research and labs.
Mythos 5 carries fewer restrictions and is being offered to a limited group of trusted organisations, mainly through Project Glasswing, which focuses on cybersecurity applications. Anthropic said access would be expanded gradually through a trust-based programme.
The split release reflects a wider trend among AI companies, which are introducing different levels of access for businesses and consumers as concerns over safety and misuse continue to grow.
OpenAI, Google and xAI have also adopted staged releases and restrictions around some capabilities, particularly in areas considered sensitive. AI developers have faced increasing scrutiny from governments, researchers and businesses over the risks posed by more capable models.
Anthropic’s latest release has also drawn debate within the AI community. Some researchers criticised an earlier policy that reduced the model’s performance on certain AI research tasks without informing users. Following criticism, the company changed its approach and said users would be notified when requests are redirected to a less capable model.
The launch comes about two months after Anthropic introduced a preview version of the Mythos model, which was made available to a small number of organisations because of concerns linked to its cybersecurity capabilities.
Founded in 2021 by former OpenAI executives, Anthropic has emerged as one of the companies competing in the market for advanced AI systems. The company has received backing from Amazon and Google and has developed a series of Claude models focused on AI safety and research.
The latest release comes as AI companies face pressure to balance access to more capable systems with concerns over security, misuse and transparency.