Anthropic on Tuesday announced the general availability of Claude Fable 5, a powerful Mythos-class AI model engineered with new safeguards that specifically restrict its use in high-risk domains, including cybersecurity.
The AI giant says this marks the first time a model of this capability class has been deemed safe enough for widespread public and developer access.
While Fable 5 demonstrates good performance — surpassing prior models in software engineering, knowledge work, vision, and long-running tasks — the company prioritized safety by implementing targeted blocks.
In sensitive areas such as cybersecurity and biology, the model automatically falls back to the less capable Claude Opus 4.8 to prevent potential misuse. Early usage data indicates that at least 95% of sessions run entirely on Fable 5’s capabilities without triggering any fallback.
“The uplift from Mythos-level capabilities is valuable to many adversaries — for instance, those who could financially gain from cyberattacks — and we therefore expect them to be motivated to try to circumvent our safety measures,” Anthropic noted.
The company emphasized the rigor of its safety measures. It conducted extensive internal red-teaming of its classifiers, followed by an external bug bounty program spanning over 1,000 hours that yielded no universal jailbreaks.
Independent external red-teaming also failed to uncover critical bypasses, underscoring the robustness of the safeguards against adversarial attempts to achieve restricted outputs.
Project Glasswing partners gain upgraded Mythos 5
Anthropic also announced on Tuesday that trusted users, including its cybersecurity partners in Project Glasswing, are being upgraded from Claude Mythos Preview to Claude Mythos 5.
The company plans to gradually expand this high-privilege access through a structured trusted-access program.
Anthropic announced recently that it’s expanding Project Glasswing to add roughly 150 new organizations.
The AI giant has not listed the new additions, but several cybersecurity and tech companies have since announced their participation in the project, including Dragos, Tenable, TrendAI (Trend Micro), Netskope, BeyondTrust, Rubrik, BT, Intercontinental Exchange, and Hitachi.
Both Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, with the former available immediately via the Claude API for developers.
Related: Claude Mythos Turns N-Days Into N-Hours With Rapid Exploit Creation
Related: New Platform Uses Cryptographic Invisibility to Protect AI-Built Applications
