GPT-5.2: AI That Outperforms Human Experts at Work

OpenAI just dropped GPT-5.2, their most powerful model yet for professional knowledge work. The big news? It actually beats or ties top human professionals on 70.9% of comparisons in knowledge work tasks across 44 occupations — think making spreadsheets, presentations, and code. And it does this at 11x the speed and less than 1% the cost of hiring experts. Enterprise users already report saving 40-60 minutes a day on average, with power users saving over 10 hours weekly. The model is better at everything: creating spreadsheets, building presentations, writing code, understanding images, handling long contexts, and using tools for complex multi-step projects. Companies like Notion, Shopify, and Zoom have already seen state-of-the-art results in reasoning and tool-calling. It’s rolling out today to paid ChatGPT plans and available to all developers via API. Why it matters: This marks a turning point where AI doesn’t just assist professionals — it matches or exceeds their output quality at a fraction of the cost, fundamentally changing the economics of knowledge work.