OpenAI released the updated GPT-5.2 model.

OpenAI has announced the launch of GPT-5.2 — an updated version of its flagship language model, designed for complex professional tasks, data analysis, and programming.

GPT-5.2 demonstrates a noticeable improvement in quality compared to previous versions. The model handles longer contexts better, follows instructions more accurately, and makes fewer factual errors. Reasoning, code processing, and image analysis skills, including charts and diagrams, have also been enhanced.

GPT-5.2 is available in several variants. The base version is intended for everyday tasks, the deep reasoning version is designed for complex analysis and logical chains, and the most powerful configuration is tailored for professional use in research and engineering environments.

OpenAI has paid special attention to safety. The model features updated mechanisms for handling sensitive content and a reduced rate of so-called "hallucinations" — confident but incorrect responses.

GPT-5.2 is already available via API for developers. In ChatGPT, the model will be rolled out gradually and will become available to users of paid plans, including Plus, Pro, Business, and Enterprise.

Source: OpenAI