58%
What it means
Marketers have long tracked share of voice: your share of the attention in your category. Share of model applies the same idea to AI: of all the brand mentions an AI model gives for your category, how many are you?
Why "per model" matters
Different AI engines recommend different brands. In our study, ChatGPT and Gemini shared only 58% of their recommendations, and 16% of recommendations came from one engine only.
So your share can be strong on one engine and weak on another. Tracking it per engine shows where to focus.
How it's calculated
Count brand mentions across many answers on one engine. Divide your mentions by the total. Do the same for each engine.
It's the same maths as AI share of voice, split by model.
Example
An accounting tool has a 14% share on ChatGPT and 4% on Gemini. Gemini leans on fresh "2026" lists that don't include it.
It focuses on getting onto those lists, and its Gemini share rises to 9% in a month.
Frequently asked questions
Is share of model a standard metric?
It's a popular name, not a standard. Different tools calculate it in slightly different ways.
Should I track models or products?
Products, like ChatGPT and Google AI Mode. That's what buyers use, and each adds its own search on top of the model.
How many answers do I need?
Enough to smooth out daily changes: dozens of questions, every day, for at least a week.