OpenAI has launched GPT-5.2, its newest frontier model series for professional work and long-running agents.
The model improves on GPT-5.1 in general intelligence, coding and long-context handling and is designed to handle more of a project end-to-end, from building spreadsheets and presentations to writing code and using tools for complex multi-step tasks.
OpenAI said GPT-5.2 sets a new state of the art on its GDPval benchmark for well-specified knowledge-work tasks across 44 occupations, beating or tying human experts on about 71% of comparisons.
In ChatGPT, GPT-5.2 Instant, Thinking and Pro are starting to roll out to paid plans, while GPT-5.1 will remain available to subscribers for three months before being retired. In the API, GPT-5.2 is offered as gpt-5.2, gpt-5.2-chat-latest and gpt-5.2-pro, and OpenAI said it has no current plans to deprecate GPT-5.1, GPT-5 or GPT-4.1.
The launch follows what Reuters reported as an internal “code red” from CEO Sam Altman in early December that paused non-core projects and redirected teams to accelerate GPT-5.2 in response to Google’s Gemini 3.
Google describes Gemini 3 as its “most intelligent model,” with state-of-the-art reasoning and multimodal capabilities, now available across the Gemini app, AI Studio, Vertex AI and AI Mode in Search.
GPT-5.2’s release alongside Gemini 3 underlines how quickly the top models are evolving, while both vendors keep earlier generations in service to support existing applications and give enterprises time to test new capabilities before switching.