THE SCOREBOARD FOR WHAT'S NEXTAI + QUANTUM, SCORED. COMPARED. EXPLAINED.
AIQUANTUMSCORE.
← AI & Quantum News

AI / NEWS BRIEF

OpenAI launches GPT-6 Cyber behind restricted access

The GPT-6 Astra variant is limited to approved Daybreak Red customers, with higher cyber capability classifications, organizational controls and premium API pricing.

By AIQuantumScore · · Event 2026-11-05

A specialized model with gated access

OpenAI has released GPT-6 Cyber, a cyber-specialized GPT-6 Astra variant for authorized vulnerability discovery, exploit development and web penetration testing. Access is restricted through Daybreak Red and requires identity verification, use restrictions, single sign-on, multi-factor authentication and API IP allowlisting. OpenAI's system card classifies the model as Critical for cybersecurity, High for biological and chemical capability, and below High for AI self-improvement.

Security results are substantial but company-reported

OpenAI says its researchers used the model to produce reproducible evidence for 81 previously unknown Firefox vulnerabilities that Mozilla classified as high severity. It also reports reproducing 1,724 Linux-kernel vulnerabilities and developing working exploits for 948, with 304 critical findings disclosed to maintainers alongside candidate patches. Reuters independently confirmed the restricted launch and reported the headline results, but the experiments and aggregate counts have not been independently reproduced; they should be read as OpenAI's claims rather than neutral benchmark results.

Capability comes with a premium and tighter controls

OUR ANALYSIS

The API price is $20 per million input tokens and $100 per million output tokens, with cached input at $2 per million. That places GPT-6 Cyber above general-purpose models in cost and outside normal self-serve availability. The release matters because it moves advanced exploit-development capability into real customer workflows, while the narrow approval process and monitoring acknowledge that the same capability raises unusually high misuse risk.

More to follow