OpenAI GPT-5.5 launch: agentic coding lead + Codex guardian + clinician tool [climaxing] [climaxing]
Key Questions
What are the key features of OpenAI's GPT-5.5 launch?
GPT-5.5 is now live in ChatGPT and Codex, topping models like Opus and Gemini with SWE-Bench at 58.6% and Terminal at 82.7%. It includes a guardian agent for safety evaluations on actions, tests, and builds. Additional perks include cheaper tokens, GB200 NVL72 VMs, and an imminent API.
What is ChatGPT for Clinicians?
ChatGPT for Clinicians is a free version of ChatGPT designed for U.S. doctors and pharmacists, compliant with HIPAA standards. It outperforms GPT-5.4 and humans on HealthBench benchmarks. It supports clinical tasks securely.
What is the guardian agent in Codex?
The guardian agent in Codex performs auto-reviews to evaluate the safety of proposed actions, tests, and builds. It is now live and enhances safety evaluations. This feature helps prevent unsafe code or operations.
Live in ChatGPT/Codex; tops Opus/Gemini (SWE-Bench 58.6%, Terminal 82.7%); guardian agent for safety evals on actions/tests/builds. Free ChatGPT for Clinicians (US docs/pharmacists, HIPAA, HealthBench beats GPT-5.4/humans). Cheaper tokens, GB200 NVL72 VMs, API imminent.