OpenAI launches GPT-5.2 with xhigh reasoning level and automatic context compression
GPT-5.2 adds an xhigh reasoning level and automatic context compression for long sessions, released through the API and in ChatGPT for general users too.
OpenAI has announced GPT-5.2, a new version in the GPT-5 line, with two notable changes: the xhigh reasoning level and an automatic context compression mechanism. The model is deployed through the API and in the ChatGPT application, meaning general users and development teams reach the same version.
The xhigh level sits at the highest step of the model’s reasoning scale, suited to problems that require multi-step analysis before answering. The automatic context compression mechanism handles situations where a working session runs long and accumulated information exceeds the context window limit, helping the conversation continue instead of having to be split or started again.
For enterprises, these two capabilities bear directly on how AI usage processes are designed. Tasks such as searching long documents, reviewing contracts or analysing logs across many steps can be handed to the model in a single session, reducing manual data-splitting work.
The point to note is that a high reasoning level usually comes with greater cost and response time, so it is not suitable to be enabled by default for every request. When older information is condensed, the possibility of losing detail is real, so processes requiring high accuracy should still include a step that checks against the original source.
Source: openai.com
