OpenAI released GPT-5.5-Cyber on June 22 as part of its Daybreak cyber defense program, achieving 85.6% on the CyberGym benchmark, slightly ahead of Anthropic's Mythos 5 at 83.8%. However, Mythos 5 and Fable 5 were taken offline on June 12 after the U.S. government issued an emergency export control directive citing national security risks due to a jailbreak vulnerability. Anthropic had no reliable way to verify user nationality and disabled both models globally. OpenAI's model, while also restricted to verified security professionals, was cleared with federal agencies before launch, including the Center for AI Standards and Innovation. OpenAI has expanded partnerships with 28 security firms and signed cybersecurity agreements with multiple nations and EU institutions. Anthropic faces additional backlash over a hidden filter in Fable 5 that degraded outputs for suspected competitors. The two-point benchmark gap is overshadowed by the regulatory and strategic differences, with GPT-5.5-Cyber remaining operational while Mythos remains suspended.

Key facts

• OpenAI's GPT-5.5-Cyber scored 85.6% on CyberGym, beating Anthropic's Mythos 5 at 83.8%.
• U.S. government banned Mythos 5 and Fable 5 on June 12 over national security jailbreak risks.
• OpenAI pre-cleared its model with federal agencies before launch, avoiding suspension.
• OpenAI signed cybersecurity partnerships with 7 countries and EU agencies.
• 28 security firms including CrowdStrike and Cloudflare integrated GPT-5.5-Cyber into products.