Anthropic Launches Claude Fable 5: First Public Mythos-Tier AI Model

Anthropic has officially launched Claude Fable 5, its first publicly available Mythos-tier model, stating that it has passed safety evaluations for general use. According to Anthropic, Fable 5 surpasses all previously released public models, demonstrating "exceptional performance" in software engineering, knowledge work, visual recognition, and scientific research. To mitigate potential risks, the model implements conservative safeguards on queries related to cybersecurity, chemistry, and biology—most such questions will be answered by Opus 4.8 rather than Fable 5 directly.

Meanwhile, Anthropic has deployed Mythos 5 through the Project Glasswing initiative, an upgrade to the previous Claude Mythos Preview. The company claims Mythos 5 has "the world's strongest cybersecurity capabilities" and plans to gradually expand access through a broader trust-based access program. Users who already have Mythos Preview access—including Apple—can now use Mythos 5. Additionally, Anthropic released the latest model Claude Opus 4.8 on the same day.

Outlook and Analysis: The launch of Fable 5 marks a key step for Anthropic in balancing capability and safety—offering a high-performance model to general users while limiting high-risk domains through a tiered response mechanism. The targeted release of Mythos 5 further strengthens the company's competitiveness in specialized fields like cybersecurity. As the Model Glasswing program expands, more enterprise customers may gain access to these cutting-edge AI capabilities in the future.