GPT-5.5 Cyber is OpenAI's latest specialised model for cybersecurity, and it is being positioned as a powerful new shield for critical digital infrastructure.
OpenAI chief executive Sam Altman has teased an imminent rollout of GPT-5.5 Cyber, the successor to the recently released GPT-5.4 Cyber model.
In a post on X, he said the system will first be offered to "critical cyber defenders" responsible for sectors such as energy, finance and key digital services, under a tightly controlled access programme.
The company is also inviting selected users to a GPT-5.5 event at its San Francisco headquarters on 5 May, where it is expected to showcase early capabilities of the new cyber-focused model.
OpenAI already classifies the broader GPT-5.5 family as having "high" cybersecurity capability under its internal preparedness framework, though still below the "critical" threshold reserved for models able to autonomously generate severe zero-day exploits at scale.
Earlier GPT-5.4 Cyber variants were tuned to help security teams reverse-engineer binaries, scan compiled software for malware and vulnerabilities, and assess resilience even without access to source code.
GPT-5.5 Cyber is expected to build on that work, pairing stronger reasoning with more permissive defensive tooling, while remaining behind OpenAI's Trusted Access for Cyber gate to limit misuse by attackers.
The model is widely seen as a direct answer to Anthropic's Claude Mythos, a frontier AI system that internal red-team tests found to be highly skilled at cybersecurity and hacking tasks.
Anthropic has restricted Mythos to a small set of vetted technology partners after it reportedly identified thousands of serious vulnerabilities across major operating systems and browsers during evaluation.
OpenAI's more gradual, tiered rollout of GPT-5.5 Cyber suggests a similar recognition that advanced cyber-capable AI must be tightly governed even as it is used to strengthen global defences.
With GPT-5.5 Cyber, OpenAI is signalling that AI will increasingly sit at the heart of digital defence, automating work that once demanded rare specialist skills.
Yet by keeping the model behind identity- and trust-based access controls, the company also acknowledges that the same capabilities could accelerate offensive operations if they leak into the wrong hands.
As rivals such as Anthropic push ahead with Mythos, governments and companies now face an uncomfortable balance: embracing these powerful tools to secure their systems, while racing to ensure they do not inadvertently help to break them