GPT‑6 Cyber - OpenAI’s cyber model, with vetted access for defenders

by•
GPT-6 Cyber is OpenAI’s cybersecurity model, which per OpenAI pairs stronger security research with better alignment. It powered OpenAI’s Linux kernel research. Rated Critical in cybersecurity, it starts with a small group of vetted Daybreak Red customers, with identity checks, MFA and an application required. OpenAI says broader access is expected over the coming weeks. No public pricing yet.

Add a comment

Replies

Best

GPT-6 Cyber is OpenAI’s new cybersecurity model, and it’s built for vetted security defenders, not for general use.

OpenAI describes it as a major advance in security research that pairs stronger capabilities with better model alignment, meaning it’s meant to be capable at the work while staying within safe limits. It’s also the model behind the Linux kernel research OpenAI announced.

It’s the newest model in OpenAI’s cyber line, which also includes GPT-5.4 Cyber, GPT-5.5 Cyber and GPT-5.6 Cyber.

Access is deliberately limited. It starts with a small group of customers in Daybreak Red, OpenAI’s application-only advanced cyber program, and broader access is expected over the coming weeks. Joining means going through identity verification and strong account security checks, so it isn’t something you can just sign up for today.

OpenAI hasn’t published pricing, and there’s no public sign-up for the full model yet.

Good fit if you work in security research, vulnerability management or defense at an organization that can apply to Daybreak, and you want to follow how frontier models are being used for finding and fixing software flaws.

Learn more:

I'm obviously not in the vetted Daybreak Red group, but the Linux kernel research part caught my eye — that's a real-world test, not just a benchmark score. Curious whether that security work ends up in the models the rest of us use day to day, or if this track is meant to stay separate for defenders.