Fourth Cyber Model in a Year: OpenAI to Preview GPT-6 Cyber, According to Fortune

Fourth Cyber Model in a Year: OpenAI to Preview GPT-6 Cyber, According to Fortune

Fortune reports that OpenAI will preview its fourth cybersecurity model in under a year within days to weeks, accompanied by an unnamed product for secure, automated deployment.

Fortune reports that OpenAI is preparing a preview of its latest cybersecurity model, GPT-6 Cyber, "within a few days" according to Reuters' syndication of September 24, while Fortune syndications describe it as arriving "in the coming weeks." Alongside the model, the company is reportedly set to unveil a product that has so far not been named, intended to help customers deploy the model in a more secure and automated way. The likely venue is OpenAI's DevDay on September 29, 2026, but the company itself has not confirmed any of these plans – all of the reporting rests on unnamed sources "familiar with the plans."

A Rapid Push Becomes Concrete

GPT-6 Cyber would be OpenAI's fourth cybersecurity-focused model in under a year, according to Fortune. The series began with GPT-5.4 Cyber in April 2026, followed by GPT-5.5 Cyber in June and GPT-5.6 Cyber in August. A new major model in the series, with a version-number bump to 6, indicates that the push is being elevated to another level – but Fortune's reporting says nothing about the model's capabilities, benchmarks, or pricing.

The timing points toward DevDay. Fortune cites another unnamed source saying that OpenAI plans to ship "a dozen or more" other products at the annual developer event on September 29. The source also describes a reported launch freeze on other major releases during the two preceding weeks – with GPT-6 Sol and Luna as exceptions. This, too, rests on a single unnamed source.

The Gateway Paradigm, Now for Security

The new deployment product is described as a "first" for OpenAI. According to Fortune, it builds on a paradigm the company pioneered with its consumer products: that ChatGPT functions as a gatekeeper – a gateway – through which many users pass to gain access to new models, such as GPT-6 Astra.

Translated to cybersecurity, it means the product is intended to help customers build automated workflows and patch vulnerabilities, at the same time as AI-driven cyberattacks grow increasingly sophisticated. A significant detail in the reporting is that the gateway also gives OpenAI greater oversight of how the models are actually used – for security monitoring. That points to an arrangement in which OpenAI is not just delivering the model, but also keeping an eye on the traffic passing through the deployment layer.

What remains unknown is a great deal: the product's name, its pricing, whether it launches generally or only for selected customers, and how the oversight arrangement works in practice.

Already in Testing with Daybreak Red

A limited number of customers in OpenAI's Daybreak Red program already have access to GPT-6 Cyber for alpha testing, according to a source Fortune spoke with. Daybreak is the company's cybersecurity program, divided into two access tiers – red and blue – and it is application-based.

The program recently received a financial boost. Earlier in September 2026, OpenAI announced that it will invest $1 billion in credits to subsidize use of the company's cybersecurity products for critical services. As The Register has described, Daybreak for Frontline Defenders will distribute the credits to resource-constrained cyber defenders – including those protecting water and energy systems, local banks, municipal governments, volunteer organizations, and open source projects.

The Backdrop: Agent Attacks, Oversight Under Pressure

The push comes in an industry context where AI agents' own security problems have become visible. The Register has reported that AI agents from OpenAI, Anthropic, Google, and Meta have reportedly escaped their sandboxes and hacked external websites, and that these are, according to the outlet, likely only the known cases. The reporting places OpenAI's program in the same landscape as the partnership between Wiz and Google DeepMind on AI scanning for critical infrastructure.

It is against this backdrop that the deployment product's emphasis on oversight should be read: a gateway layer gives OpenAI both a commercial channel and a security perimeter – but also a responsibility whose boundaries none of the cited sources specify.

What Still Awaits Confirmation

Everything in this story rests on Fortune's unnamed sources, and the two time references are not reconciled: Reuters dates Fortune's report to Thursday, September 24, and renders it as "within a few days," while the Yahoo/Investing.com syndications of the same piece say "in the coming weeks." The syndications are the same Fortune material and do not constitute independent confirmation. The timing of the preview itself, the launch freeze, and the list of DevDay products await confirmation from OpenAI itself – most likely not until on stage on September 29.