News › AI & LLM

AI & LLM

OpenAI launches GPT-6 Cyber: a security model rated “Critical”, priced at $20/$100 per million tokens

2026-11-05 · Source: VentureBeat

OpenAI launches GPT-6 Cyber: a security model rated “Critical”, priced at $20/$100 per million tokens
Image: Reuters / The Indian Express

According to OpenAI, VentureBeat and Reuters, OpenAI released GPT-6 Cyber, a model specialised for security research, on Nov 5, 2026. It initially goes to a group of approved customers of Daybreak Red, OpenAI's access tier for advanced, authorised security work such as penetration testing and exploit development; access will extend to all approved Daybreak Red customers over the coming weeks, while GPT-5.6 Cyber opens to all Daybreak Red customers the same day. OpenAI describes GPT-6 Cyber as a variant of GPT-6 Astra with additional training for security investigations: reinforcement learning rewards reproducing crashes and building exploits alongside behaviour that respects authorised scope. According to OpenAI, the model is rated “Critical” for cybersecurity capability under its Preparedness Framework (GPT-5.6 Cyber: “High”). Figures published by OpenAI: 48.5% on ExploitGym (GPT-5.6 Cyber: 32.1%); in web-application testing without source code the model found 40.3% of vulnerabilities (GPT-6 Astra: 32%); on RemoteExploitBench, which covers known vulnerabilities in applications such as GitLab, PostgreSQL and Jenkins, it completed 95% of tasks. OpenAI's researchers used the model to reproduce 1,724 Linux kernel vulnerabilities and build working exploits for 948 (588 of them previously unknown, according to Reuters), and disclosed 304 critical findings to Linux maintainers with proposed patches; for Firefox, OpenAI reports 81 previously unknown vulnerabilities that Mozilla classified as high severity, with Mozilla having begun to release fixes. In the ExploitGym Honeypot test (the model must recognise that vulnerable services placed next to its assigned target are out of scope), without deployment safety prompts GPT-6 Cyber avoided unauthorised attacks in 99.93% of runs (GPT-5.6 Cyber: 94.49%). Listed API prices are $20 per million input tokens and $100 per million output tokens (cached input: $2), higher than GPT-5.6 Cyber ($12.50 and $75).

Source: VentureBeat

Need an LLM chatbot or agent for your business?

Describe your need briefly; you will get clarifying questions and a scoped proposal before any work starts.

Send a request See the service

Related news