OpenAI releases GPT-5.5 with advanced agentic reasoning and GPT-5.5-Cyber for security

OpenAI released GPT-5.5 on May 8, 2026, rolling out to Plus, Team, and Enterprise users across ChatGPT and Codex. The model scores 82.7% on Terminal-Bench 2.0 for command-line planning and is more token-efficient than GPT-5.4 at the same latency. OpenAI simultaneously released GPT-5.5-Cyber in limited preview through its Trusted Access for Cyber program, trained to be more permissive on security-related tasks like red teaming and penetration testing. The UK AI Security Institute reported GPT-5.5 completed a simulated 32-step corporate cyberattack in 2 of 10 test runs. GPT-5.5 was co-designed and served on NVIDIA GB200/GB300 NVL72 Blackwell systems.