ChatGPT-5.2 Is Here – Here's What The New Model Can Do

ChatGPT 5.2 has arrived, and if you're expecting a raft of fun features, prepare to be disappointed: 5.2 is all about "professional knowledge work". While 5.1 gave us new personality options for ChatGPT,introducing tones like "Candid", "Quirky", and "Nerdy", this upgrade wants to make us more efficient at the office. It's not going to be anything as exciting as that time ChatGPT 4o was released, and everyone started producing Studio Ghibli-style images of themselves.

The new developments are intended to enable people to pass on more of their gruntwork to AI tools. According to OpenAI's press release, some ChatGPT users are already saving ten hours a week by delegating tasks to AI. And with 5.2, OpenAI says we'll be able to "unlock even more economic value". ChatGPT 5.2 can create spreadsheets, put together presentations, and handle multi-step work projects. Pretty soon, we might not need humans in the office at all. It'll be all AIs whipping up spreadsheets and presenting them to each other in the boardroom. According to OpenAI, ChatGPT can now perform better than most human experts. In tests, 5.2 outperformed 70% of business professionals although this was limited to spreadsheet- and presentation-related tasks, so we're probably not close to AI superintelligence just yet. 

Users on both paid and free plans can access ChatGPT 5.2 right now, but free users only get a limited number of goes. There are three versions:  Instant, Thinking, and Pro. Instant provides quick answers for straightforward questions, Thinking takes care of more complex tasks like coding, long documents, and step-by-step reasoning, and Pro promises accurate answers for challenging problems but takes the longest to reply. The Pro version is only available on the pricier paid plans.

OpenAI's press release presents a raft of metrics that demonstrate how well 5.2 performs across benchmarks, besting 5.1 in knowledge work tasks, software engineering, advanced mathematics, and abstract reasoning. But even if it's better than its predecessor, ChatGPT isn't the only AI tool in town. When you compare ChatGPT 5.2 to other large language models (LLMs), it may be outperforming its rivals in reasoning tasks — but not by much. It scores 92.4% on the GPQA Diamond benchmark — which tests AI performance on expert-level science questions — but Google's Gemini 3 Pro isn't far behind with 91.9%. On Agentic Coding and Visual Reasoning benchmarks, ChatGPT lags behind Anthropic's Claude. But although it might not be number one in everything, it still ranks in the top three across the major categories.

AI tools — including ChatGPT — are advancing at a phenomenal rate. This new upgrade suggests that it isn't slowing down yet. So what does all this mean for us? Is it another step towards us all being replaced at work by AI? While OpenAI might frame it as "helping users tackle more complex tasks with greater polish," if AI can do the jobs of everyone but the top 30% of white-collar employees, this is going to benefit CEOs and managers more than the rest of us. Time will tell if 5.2 lives up to its hype — OpenAI has a history of overpromising — but if your day job currently involves making a lot of spreadsheets, you might be feeling a bit nervous right now.