OpenAI's GDPval Benchmark Evaluates AI in Real-World Jobs

OpenAI has introduced a new benchmark called GDPval to evaluate how its AI models, including GPT-5, perform compared to human professionals across a wide range of industries. In a company announcement, OpenAI detailed that GDPval spans 44 occupations within nine key industries contributing significantly to the U.S. GDP, such as healthcare, finance, and manufacturing.

The GDPval benchmark is designed to test AI models on real-world tasks that are economically valuable. It involves tasks like creating legal briefs, engineering blueprints, and customer support conversations, which are evaluated by experienced professionals. The initial results show that GPT-5 and Anthropic's Claude Opus 4.1 are approaching the quality of work produced by industry experts, with GPT-5 being ranked as better than or on par with experts 40.6% of the time.

OpenAI acknowledges that GDPval is an early step and plans to expand the benchmark to include more interactive workflows and context-rich tasks. This initiative aims to provide a clearer picture of how AI can support professionals in their day-to-day work, potentially allowing them to focus on more meaningful tasks.