GPT-5.2 is OpenAI's December 2025 flagship for professional knowledge work and long-running agents, positioned as the most capable series in the GPT family for tasks that involve sustained reasoning, multi-step planning, and tool use. The official launch frames it as a step up from prior GPT models in creating spreadsheets, building presentations, writing code, perceiving images, handling long contexts, and orchestrating complex agentic workflows, making it a strong fit for enterprise pipelines that need a single model to cover research, drafting, and execution stages.

In benchmark terms, the Thinking variant posts notable gains over the previous generation, leading GDPval with a 70.9% win-or-tie rate on knowledge-work tasks across 44 occupations versus 38.8% for GPT-5.1 Thinking, and showing double-digit improvements on reasoning suites such as ARC-AGI-2 Verified (52.9% versus 17.6%) and GPQA Diamond (92.4% versus 88.1%). It also reports state-of-the-art results on software-engineering evaluations like SWE-Bench Pro (55.6%) and SWE-bench Verified (80.0%), suggesting that for buyers choosing a frontier general-purpose model, GPT-5.2 is best suited to workloads that combine document understanding, code generation, and extended multi-step reasoning rather than narrow single-turn chat.