OpenAI limits GPT-5.5 cyber access

- OpenAI launched GPT-5.5‑Cyber in a limited preview and expanded its “Trusted Access for Cyber” program, restricting the model to verified defenders. - At the consumer level, ChatGPT's Instant mode can now pick between GPT-5.3 Instant and GPT-5.5 Thinking, switching to deeper reasoning for harder requests. - The rollout shows capability gains paired with tighter gating and updated enterprise/education model limits in OpenAI's docs. (pulse2.com) (help.openai.com)