OpenAI has unveiled GPT-5.2, its most advanced model yet, with improvements in spreadsheets, presentations, coding, vision, long-context understanding, tool use and multi-step project execution.
GPT-5.2 is “better at executing complex, real-world tasks end-to-end than any previous model,” the company said in a blog post.
OpenAI says the average ChatGPT Enterprise user saves 40 to 60 minutes per day, with heavy users saving more than 10 hours per week.
GPT-5.2 comes in three flavors: Instant, Thinking and Pro. They are available in ChatGPT paid plans and are also offered in the API now, with new reasoning-effort options and published token pricing.
OpenAI said GPT-5.2 Thinking performed well on GDPval, an evaluation of well-specified knowledge work across 44 occupations: It beats or ties industry professionals 70.9% of the time. OpenAI claims GPT-5.2 Thinking is more than 11x faster than human experts and is less than 1% of their cost. GPT-5.2 Pro reaches 74.1%.
GPT-5.2 Thinking also posted gains on coding benchmarks (55.6% on SWE-Bench Pro and 80.0% on SWE-bench Verified), science and math (92.4% on GPQA Diamond; 100% on AIME 2025; higher FrontierMath scores), and abstract reasoning (86.2% on ARC-AGI-1 and 52.9% on ARC-AGI-2).