Apify releases real-time price grounding tutorial

Apify released a tutorial on how to do price grounding for LLMs with real-time data and build a Python agent. The tutorial requires Python 3.11 to 3.14, an Apify account (the Free plan requires no credit card), and an OpenAI API key; the Free plan includes $5 of usage per month, which Apify says is enough for hundreds of real-time price queries. The tests cited in the tutorial show that without tools, GPT-6 Luna and Claude Opus 5.5 each answered only 2 real-time prices correctly out of 15 products, with median errors of 12% and 17%, respectively; after enabling built-in web search, exact prices rose to 9–11 and 7–9, respectively, but each run still had incorrect prices, and source links were attached. The tutorial says grounding requires real-time data, a clear status when checks fail, and prices written by code rather than by the model. Browsing-type questions use cached prices no more than 1 hour old.

AI overview based on source reports · Updated 3 days ago