OpenAI GPT-5.5 release [developing]
Key Questions
What is OpenAI's GPT-5.5 model focused on?
GPT-5.5 is a new agentic model for real work, live in ChatGPT and Codex. It handles complex goals, tools, self-checks, efficient coding/debugging, and multi-tasking.
Where is GPT-5.5 now available?
GPT-5.5 is live in ChatGPT and Codex, enabling a new class of AI for agents and real work.
What improvements does GPT-5.5 offer in efficiency?
Sam Altman praised the inference team's work for serving GPT-5.5 efficiently, with improved token usage noted in launches like GPT-5.1.
What evaluations are incoming for GPT-5.5?
Abacus evals and LiveBench are upcoming to assess the model's capabilities.
What example demonstrates GPT-5.5's capabilities?
GPT-5.5 built a full game in one shot, as featured on the release page and shared by users.
New agentic model for real work live in ChatGPT/Codex; complex goals/tools/self-checks/efficient coding/debugging/multi-task; Abacus evals/LiveBench incoming.
Sources (6)
Updated Apr 24, 2026