GPT-5.5

GPT-5.5 is a frontier-class system positioned by its developers as their smartest and most intuitive model to date, designed to take messy, multi-part objectives and carry them through to completion on its own. Rather than waiting for carefully staged instructions, it plans a workflow, selects tools, checks intermediate results, navigates ambiguity, and continues iterating until a task is finished. According to OpenAI's official release notes, this autonomy is paired with strong general reasoning, so the same model can write and debug code, research online, analyze datasets, author documents and spreadsheets, and operate other software as part of a single agentic loop.

The model is tuned for real-world serving efficiency as well as raw capability, delivering a higher level of intelligence while matching the per-token latency of the previous generation in production. It uses substantially fewer tokens to complete equivalent Codex coding tasks, which makes long, tool-heavy workflows cheaper to run without sacrificing quality. OpenAI highlights especially large gains in agentic coding, computer use, knowledge work, and early-stage scientific research, areas where progress depends on sustained reasoning across that quick-info value and taking actions over time. GPT-5.5 Pro accompanies the base release in the API for users who need the highest-tier reasoning depth.