Anthropic Unveils Claude Fable 5 and Mythos 5 Models With Strict Cyber Safety Filters
Anthropic releases Claude Fable 5 as its most capable general-use AI model, featuring state-of-the-art performance across various benchmarks. The company also introduces Claude Mythos 5 for select cyberdefenders, though access to both models quickly faces a temporary suspension.
Anthropic officially launches Claude Fable 5, a highly advanced AI model that outperforms all previous models on nearly every tested capability benchmark. The model shows exceptional skill in software engineering, scientific research, knowledge work, and vision, with its advantages growing significantly as tasks become longer and more complex. Fable 5 operates at a competitive price point of $10 per million input tokens and $50 per million output tokens, making it less than half the cost of the previous Claude Mythos Preview.
Because the model possesses immense capabilities that malicious actors could exploit in areas like cybersecurity, Anthropic implements conservative safeguards to ensure safe general use. These safety filters automatically redirect certain sensitive queries to Claude Opus 4.8, triggering in fewer than five percent of sessions and occasionally catching harmless requests. For specialized government and infrastructure partners, Anthropic offers Claude Mythos 5, which removes these specific restrictions to provide the world's strongest cybersecurity capabilities through Project Glasswing.
Despite the impressive potential for these models to accelerate life sciences research and secure critical software, Anthropic abruptly suspends access to both Fable 5 and Mythos 5 just days after the initial announcement. The company apologizes for the sudden disruption to its customers and states that it is actively working to restore availability as quickly as possible.