OpenAI announces GPT-5.5 with a focus on coding and efficiency

In short: OpenAI says its new GPT-5.5 AI model is more efficient and better at coding, and it starts rolling out to paid users this week.

What happened

OpenAI announced GPT-5.5, a new version of the AI model behind ChatGPT. The company released GPT-5.4 last month, and now says GPT-5.5 is its most capable model so far and easier to use.

OpenAI says GPT-5.5 does especially well at coding tasks, like writing code and debugging it (finding and fixing mistakes, like proofreading but for software). It also says the model can handle “messy, multi-part” work, plan the steps, use different tools, check its own work, and keep going even when the instructions are not perfectly clear.

The company lists other tasks it is aiming for, including doing research online, and creating spreadsheets and documents. GPT-5.5 will roll out starting Thursday to ChatGPT Plus, Pro, Business, and Enterprise plans, and to Codex. Codex is OpenAI’s coding product, built to help developers write and fix software.

OpenAI also says GPT-5.5 uses fewer “tokens” in Codex. Tokens are small chunks of text that AI systems count as they work (like a taxi meter, but for words).

Why it matters

Many people are starting to use AI for everyday office work, not just chat. If GPT-5.5 really needs less step by step guidance, it could make AI feel more like a helpful assistant that can complete a whole errand, rather than a tool you must supervise closely.

Source: The Verge AI