OpenAI released GPT-5.2 in three variants—Instant, Thinking, and Pro—as part of a broader push to deepen its enterprise capabilities and core user experience.
During a press briefing, OpenAI executives said 5.2 is designed to unlock greater economic value, with capabilities spanning spreadsheets, presentations, coding, image understanding, and multi-step project linking.
GPT-5.2 incorporates a 400,000-token context window and a knowledge cutoff of August 31, 2025. Subscriptions for ChatGPT are being upgraded to GPT-5.2, and API access is available to developers. GPT-5.1 remains available for three months in a legacy model path, with pricing at $1.75 per million input tokens for the standard model.
OpenAI’s release follows Google’s Gemini 3 performance surge, which spurred Sam Altman to issue an internal “code red” focused on core ChatGPT improvements. Google’s Gemini app reportedly has hundreds of millions of users; OpenAI claims 800 million weekly active ChatGPT users, a figure that underscores competition in AI.
Early benchmark figures touted by OpenAI show GPT-5.2 Thinking achieving strong results on SWE-Bench Pro and GPQA Diamond benchmarks, with indications that the model matches or surpasses some human performance on GDPval tasks. The company cautions that external, independent benchmarks will take time and that real-world results may vary as user feedback comes in.
