OpenAI is lining up GPT-6 Cyber and a security product, Fortune reports
Fortune, 6:29 p.m. ET Thursday: fourth cyber model this year; Daybreak Red already in alpha; unnamed deploy-and-patch product. First said days / DevDay Sep 29; later updated to weeks. Reuters could not verify. OpenAI did not comment.
Feature2 minIn-depth
Maya Chen2 min read
- OpenAI
- GPT-6 Cyber
- DevDay
- Daybreak Red
- Cybersecurity

Photo: Getty Images / FortuneSource page
OpenAI is preparing to preview GPT-6 Cyber, a cybersecurity model, and to show a still-unnamed product that would help customers deploy it in a more secure, automated way, Fortune reported on Thursday at 6:29 p.m. Eastern, citing multiple people familiar with the plans. Reuters picked the story up at 10:58 p.m. UTC, said it could not immediately verify it, and said OpenAI had not answered a request for comment outside business hours. Fortune later updated the piece to say the launch could be within weeks, not days. The first version had pointed at DevDay in San Francisco on Tuesday, 29 September, or sooner, with a commercial launch in the coming months.
A limited set of customers in Daybreak Red, OpenAI’s application-only top cybersecurity tier, already have GPT-6 Cyber in alpha, Fortune’s sources said. Daybreak Blue is the more general, still application-only, track. The new product is meant to sit in front of the model the way ChatGPT sits in front of GPT-6 Astra: automated workflows, vulnerability patching, and more of OpenAI’s own oversight of how the weights are used. It would be the company’s fourth cyber-branded model this year, after GPT-5.4 Cyber in April, 5.5 in June and 5.6 in August. Enterprise cyber sales now sit with Dali Rajic, the chief revenue officer who joined in August. Earlier this month OpenAI said it would put $1 billion toward subsidizing cybersecurity products for critical services.
The same sources told Fortune the company froze major launches for about two weeks, aside from GPT-6 Sol and Luna, so it could stack them at DevDay. Altman had teased a run of ships on X. The cyber preview lands in the same week Australian officials described an OpenAI agent reaching a Medicare statistics portal, and after a string of agent incidents that included Hugging Face. Fortune framed the new tools as OpenAI selling defense, not as a fix for those cases. Reuters printed the days timeline and the verification caveat. Neither outlet had pricing, a capability card, or an on-the-record confirmation.


