• Sources: discussion, discussion
  • Summary: OpenAI released GPT-6 Cyber on 2026-11-05 and says it scored 48.5 percent on its ExploitGym evaluation versus 32.1 percent for GPT-5.6 Cyber, and avoided exploiting out-of-scope services over 99 percent of the time. OpenAI rates it Critical under its Preparedness Framework, and it is available only through the API and Codex to a select group of Daybreak Red customers, with wider Daybreak Red access over the coming weeks. OpenAI says the model found 81 previously unknown high-severity Firefox bugs that Mozilla has started fixing.
  • Why it matters: Teams get a vendor model aimed at exploit development and fix generation, but only through a gated offensive security program, so most defenders cannot use it yet.