Latest / Frontier models and safety
OpenAI releases GPT-6 Cyber for approved security customers
See how this story connectsOpenAI released GPT-6 Cyber on 5 November 2026, a cybersecurity-tuned model first available through the API and Codex to a select group of approved Daybreak Red customers. On OpenAI’s ExploitGym training evaluation it scored 48.5%, against 32.1% for GPT-5.6 Cyber, and it avoided exploiting out-of-scope services in over 99% of a separate evaluation. OpenAI said the model found 81 previously unknown high-severity Firefox vulnerabilities and helped develop exploits for 948 Linux kernel vulnerabilities, 588 of them previously unknown.
Why it matters
Access is limited to customers authorised for offensive security work, and OpenAI rates the model Critical, its highest cyber risk tier, under its Preparedness Framework.
- When
- 5 Nov 2026
- ID
WTA-N-2026-0740- Where
- International
- Who
- OpenAI
Summary written with AI by wheresthe.ai, from The Register and the reporting credited above. Checked against the source on 5 Nov 2026.
Source copy: saved on 5 Nov 2026 to check this story and not republished, because the text belongs to its publisher. Also held by the Internet Archive.
Cite this story
WTA-N-2026-0740: “OpenAI releases GPT-6 Cyber for approved security customers.” wheresthe.ai, 5 Nov 2026. https://wheresthe.ai/d/intl-openai-releases-gpt-6-cyber-for-approved-security-custo-2026-11/ (source: The Register).
Spotted an error? Email [email protected].