Anthropic on Tuesday released Claude Fable 5, making a Mythos-class AI model publicly available for the first time, while simultaneously launching Claude Mythos 5, a more powerful variant that remains restricted to vetted organisations under its Project Glasswing programme.
What Is Claude Fable 5
Claude Fable 5 is a Mythos-class model, placing it on par with Anthropic’s most advanced internal systems in terms of capability. It is a general-purpose AI designed to manage complex, multi-step tasks in areas like software engineering, logical reasoning, and scientific analysis. In contrast to earlier Claude versions, Fable 5 is optimized for continuous, sustained workflows rather than isolated prompts, making it function more like an agent-style system than a conventional AI chatbot.
Fable 5 targets real-world workflows including drafting long-form content, generating design concepts, producing product documentation, and assisting with code, handling multi-stage problems and visual inputs more reliably than earlier Claude versions.
How It Differs From Mythos 5
The dual release marks a shift in how Anthropic is deploying frontier AI. Instead of limiting access entirely, the company is separating capability from access, exposing a controlled version publicly while keeping full capabilities within restricted environments. Claude Mythos Preview was the first version, released under Project Glasswing focused on cybersecurity.
Claude Mythos 5 is the updated version of that system, continuing to operate under restricted access. Claude Fable 5 is the public-facing version, sharing the same architecture and capability level as Mythos 5, but with safeguards that limit behaviour in high-risk domains.
Anthropic said Fable 5 includes strict safety controls in areas such as cybersecurity, biology, and chemistry. Requests involving restricted activities are blocked or redirected to a less capable model. The company also conducted extensive red-team and bug bounty testing before launch, reporting no reliable jailbreaks.
According to Anthropic, Fable 5 matches Mythos 5 on coding, reasoning, and knowledge-based tasks, outperforming several earlier models and competitors on select benchmarks. The model is available globally through the Claude API and Enterprise plans, while access to the more powerful Mythos 5 remains limited to approved organisations under Project Glasswing.
The launch comes as Anthropic prepares for a potential IPO, with Fable 5 serving as its first public release of a Mythos-class model and a key step toward wider commercial deployment of its most advanced AI technology.