This test is designed to show whether Open AI’s browsing tool does a better job at supplying their model GPT-5 with grounding context from a page with schema. We took the exact HTML from the original experiment here, stripped off the “experiment” from the title and header and uploaded here and here and then ran […]
