The Gemini (Vertex) API now gives limited access to grounding queries and returns the search phrases in the [‘groundingMetadata’] object. That’s exciting news, in theory, but it turns out to be messy in practice. We encountered two major issues:
Even running Gemini 3.5 Flash with a [MEDIUM] or higher “thinking” level, it’s rare to get more than 1-2 grounding queries. In many cases, you’ll see none at all. If you’ve used AI Mode much, you know it almost always runs multiple grounding queries. We were able to nudge the system instructions and push Google to generate 4-6 grounding queries. While these queries are “real” (Google actually ran the searches), it’s fair to say that we cheated a little bit. AI Mode and actual query fan-out might behave very differently.
On top of token costs, grounding queries currently cost (as of August 2026) $14 per 1,000 queries. This is not a per-prompt cost. In our case, each prompt ran about five grounding queries, so grounding data cost something like $7 per 100 prompts. This is orders of magnitude more than the token cost, and at enterprise-scale, it adds up fast. If you’re using 3rd-party tools that run on your own API keys, keep a close eye on costs.
If you liked Demystifying Grounding Queries with Real Data by Dr. Peter J. Meyers Then you'll love Miami SEO Expert
In short, an agent comes to your website, discovers what WebMCP tools you offer, reads…
GBP category: not what's technically accurate, but what's driving discovery in this market right now.…
Claneo’s 2025 State of Search study supports this. Search patterns vary by intent and age…
The SERP has turned into a crowded billboard of ads, AI features, and rich results.…
Workflow 3: Competitive analysis with chained agentsCompetitive analysis is one of my favorite AI use…
So how do I optimize my websites for agentic AI to make sure that it…