Anthropic Opens Claude Fable 5 To Public With Safeguards And Token Pricing

Anthropic opened its most powerful AI model to the general public on Tuesday, releasing Claude Fable 5 as the first widely available model from its Mythos class while restricting use in sensitive areas.

The release comes as Anthropic said Fable 5 is exceptionally good at writing and debugging software code, answering complex research questions, and analyzing images, and it is being offered at $10 per million input tokens and $50 per million output tokens.

Altitudes Magazine

Anthropic also said it is offering an unrestricted version, Claude Mythos 5, to companies and organizations that already have access to this model family, including cybersecurity partners enrolled in its Project Glasswing program.

In parallel, Anthropic said most queries about cybersecurity or biology and chemistry will be routed instead to the lower-tier model, Opus 4.8, which was made public in late May.

The company said it had identified large-scale attempts to extract its technology to train competing AI models in authoritarian countries, and these type of queries will also fall back to the less powerful model.

Anthropic said it built extensive safeguards into Fable 5 to prevent misuse in sensitive areas, including cybersecurity and certain biological research topics, and that high-risk requests are redirected to Claude Opus 4.8.

The company said it subjected the system to extensive internal and external testing to determine whether users could bypass its protections, with more than 1,000 hours of independent red-team testing and bug bounty exercises failing to uncover a universal method.

Anthropic

Dianne Penn, Anthropic’s head of product management, research and labs, told South China Morning Post that “Let’s say I’m a college student asking the model like help me find cyber vulnerabilities on X package or code. The model would refuse and Fable 5 will fall back to Opus 4.8 for a response.”

In its own announcement, Anthropic said the safeguards are tuned conservatively and “they’ll sometimes catch harmless requests, though they trigger, on average, in less than 5% of sessions.”

The public release intensifies a race among AI developers as Anthropic disclosed it had confidentially filed paperwork for an initial public offering just one week before unveiling Claude Fable 5.

France 24 reported that both Anthropic and rival OpenAI announced in the past week that they had filed IPO plans, and it said SpaceX is expected to break records with its market debut Friday.

Anthropic said it is deploying Mythos 5 in collaboration with the US government and that it is expanding access for a small group of cyberdefenders and infrastructure providers through Project Glasswing.

The company also framed the stakes in terms of misuse risk, warning that “Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage,” while adding that it is working to improve safeguards and reduce false positives as more capable models arrive.

South China Morning Post said access to the model had previously been limited to organisations like the US government under the Glasswing programme, and it described the preview earlier this year as sending “shock waves globally” with its ability to find software flaws.