OpenAI has launched GPT-5.2, a new model aimed at professionals and developers, to compete with Google's Gemini 3. GPT-5.2 is designed to excel in complex problem-solving, coding, and agent-style workflows. It is available in three versions: Instant, Thinking, and Pro, each tailored for different tasks. GPT-5.2 Instant is for quick, everyday tasks, Thinking is for deeper reasoning and coding, and Pro is for demanding scientific and technical questions.
GPT-5.2 Thinking has shown significant improvements in reasoning and accuracy, with a 30% reduction in factual errors compared to GPT-5.1. It achieved a perfect score on the AIME 2025 mathematics benchmark and outperformed Gemini 3 Pro on several metrics, including scientific and mathematical reasoning. The model also demonstrates superior performance in long-context reasoning, making it suitable for analysing extensive documents. GPT-5.2 supports the Responses API, enabling developers to build agent systems more efficiently. The new models are being rolled out to paid ChatGPT Plus, Pro, and Enterprise subscribers.
