OpenAI ChatGPT Workspace Agents Rollout
Key Questions
What is GPT-5.5 'Spud'?
GPT-5.5 'Spud' is OpenAI's latest model powering Codex team agents, workflows, orchestration, context, and safety features.
What performance does GPT-5.5 achieve on Terminal-Bench?
GPT-5.5 scores 82.7% on Terminal-Bench for coding, research, and office autonomous multi-step tasks.
What are early evals for OpenAI's new model?
Early evals for GPT-5.5 include LiveBench benchmarks, with reports from early access users confirming strong performance.
What powers ChatGPT Workspace Agents?
ChatGPT Workspace Agents rollout is powered by GPT-5.5 'Spud', enhancing agentic workflows.
What are the key improvements in GPT-5.5?
GPT-5.5 excels in coding, computer use, and deeper research, as announced by OpenAI.
GPT-5.5 'Spud' powers Codex team agents workflows HN96pts orchestration context/safety; 82.7% Terminal-Bench coding/research/office autonomous multi-step; early evals LiveBench.
Sources (4)
Updated Apr 24, 2026