Anthropic's Claude Fable 5 Hits #1 Then Gets Pulled by US Government
Anthropic's Mythos-class model goes public with safety guardrails, tops agentic benchmarks, then gets pulled by a US government export directive within days
Claude Fable 5 is Anthropic's most capable model ever made publicly available, and it had one of the most turbulent launches in AI history. Released on June 9, the model scored #1 on Artificial Analysis's GDPval-AA agentic benchmark with a score of 1932, put three Anthropic models in the top four spots, and drew immediate praise from Cursor, GitHub, Cognition, and others. Three days later, a US government export directive forced Anthropic to pull it offline for all customers worldwide.
Mythos, finally democratized
Launched as a preview in April, Mythos was initially limited to a handful of partners due to cybersecurity concerns. The model's offensive cyber capabilities were simply too powerful to release broadly. Anthropic captivated Wall Street and government officials with the unveiling of Mythos, which excels at identifying security flaws within software. The company said it did not plan to make the model generally available, and limited the rollout to a select group of companies as part of a cybersecurity initiative called Project Glasswing.
Fable 5 is the public, guardrailed sibling of Mythos. The two are not separate architectures. Anthropic states that Claude Fable 5 and Claude Mythos 5 share the same underlying model and the same published specifications. The name difference is intentional: Fable comes from the Latin fabula, meaning "that which is told,