Sam Altman's OpenAI on Thursday unveiled GPT-5.2, its most advanced AI model yet, calling it the strongest version so far for everyday professional use.
In ChatGPT, there are three versions of GPT-5.2: Instant, Thinking and Pro, according to OpenAI.
The versions are now live for paid users.
OpenAI describes GPT-5.2 Instant as the fast, everyday model meant for quick answers. It gives clearer explanations and does better with how-to questions, technical writing and translation, .
What makes GPT 5.2 better?
GPT-5.2 Thinking is designed for heavier tasks. It handles coding, long document summaries, math, logic and planning with more structure and polish.
OpenAI says this model produces deeper, more detailed responses.
GPT-5.2 Pro is the top-end version for people who need the most accurate answers, even if it takes longer.
Early testing shows fewer major mistakes and stronger performance in complex areas like programming.
Across all versions, GPT-5.2 improves reasoning, understanding long documents, using external tools and interpreting images, making it better at completing real-world tasks from start to finish.
The model makes fewer mistakes overall, with 30% fewer inaccurate responses compared with GPT-5.1 Thinking. OpenAI says this should make it more dependable for research, writing, analysis and decision-making.
Vision abilities have also improved. GPT-5.2 Thinking cuts error rates roughly in half when interpreting charts, dashboards, diagrams and software interfaces. This helps users in fields like finance, engineering, operations, design and customer support, where visuals play a key role.
Why GPT 5.2 Thinking stands out?
Among the three, GPT-5.2 Thinking stands out.
On GDPval, a benchmark that tests professional-quality work across 44 occupations, it sets a new record and becomes OpenAI’s first model to match or exceed expert-level performance.
Experts say it beats or ties human professionals in about 71 per cent of comparisons.
The new spreadsheet and presentation features in ChatGPT only work with GPT-5.2 Thinking or Pro and require a Plus, Pro, Business or Enterprise plan.
OpenAI says complex files may take a few minutes to generate.
GPT-5.2 Thinking also posts major gains in real-world software engineering.
Testers also found it much better at front-end development and complicated UI designs, including 3D elements.