OpenAI is preparing to release GPT-6 Cyber, its fourth cybersecurity-focused AI model this year, along with a new security gateway product to deploy it, according to a Fortune report cited by multiple outlets. The model could be unveiled at the company's annual DevDay event in San Francisco on September 29.

The new product will act like a gateway, similar to how ChatGPT serves users of OpenAI's consumer models. It will let enterprise customers build automated workflows and patch vulnerabilities faster, while giving OpenAI more visibility into how the model is used so it can monitor for safety risks.

A limited number of customers in OpenAI's application-only Daybreak Red program already have access to GPT-6 Cyber for alpha testing, according to the report. Earlier models GPT-5.4, GPT-5.5 and GPT-5.6 Cyber were released in April, June and August. The release comes amid industry efforts to set up an AI safety standards body by early next year.

Analysts say the launch marks OpenAI's push into enterprise cybersecurity, a growing market as AI-driven attacks escalate. What happens next: the company is expected to preview the products at DevDay on September 29, with a full launch in the coming months.