FreeLLMOpenAI launches GPT-6 Cyber, a security‑focused AI model
GPT-6 Cyber finds hundreds of new bugs, builds exploits and stays within guardrails, but is limited to select customers.
OpenAI announced GPT-6 Cyber, a variant of GPT-6 Astra trained to discover software vulnerabilities, create working exploits and suggest patches.
In a test on Firefox, the model uncovered 81 previously unknown critical flaws that Mozilla has already patched, and flagged 30 known issues as more severe.
A Linux‑kernel campaign scanned about 5,034 security‑relevant locations, confirmed 1,724 vulnerabilities and produced 948 exploits, reporting 304 critical findings to maintainers.
On the ExploitGym benchmark the model achieved a 48.5 % success rate, compared with 32.1 % for its predecessor, and a 40.3 % hit rate on source‑code‑free web‑app testing versus 32.0 % for GPT‑6 Astra.
What it means for you
For defenders, GPT-6 Cyber offers a powerful, vetted tool that can accelerate vulnerability research while staying within strict guardrails. The limited rollout means free or local models still lack comparable security capabilities, underscoring the gap between open‑source options and specialized, paid AI services.
Sources
- OpenAIs GPT-6 Cyber can both break into systems and propose how to fix them, The Decoder, 5 Nov 2026
- GPT-6 Cyber System Card - OpenAI Deployment Safety Hub, deploymentsafety.openai.com (official), 5 Nov 2026
Written automatically from the sources above, then checked by our code: every number and name must appear in them, and no sentence may copy a source. Spotted a mistake? Report it.
Which AI can you use for free today?
Every free AI model, ranked by blind human votes, with the memory and speed it needs on your own GPU.