Which AI engine is most likely to mention a given company at all?

A company trying to understand its presence in AI search usually starts by checking one assistant, often ChatGPT because it is the most familiar. This study's results suggest that is an incomplete picture, because the three engines tested do not show the same breadth of companies.

Of the 100,966 companies in the cohort, ChatGPT showed 43.9% at least once across the study's 14,951 prompts. Claude showed 40.9%. Gemini showed the narrowest range, 32.6%. That is a meaningful spread: a company could be entirely absent from Gemini's answers while still appearing regularly on ChatGPT, simply because Gemini's underlying process surfaces a smaller slice of the total field of eligible companies in any given answer.

This is a measure of breadth, not favoritism toward any particular company. It describes how many distinct companies out of the full cohort got mentioned at all on each engine, across the entire run window of August 4 to August 31, 2026. It does not mean ChatGPT is more generous to any one company, or that Gemini is stingier toward any one category. It means that, averaged over 14,951 prompts and the full company list, ChatGPT's answers collectively touched a wider swath of the cohort than Gemini's did.

Why might this happen? A few mechanisms are plausible and the study's data cannot fully separate them. One is answer length: if ChatGPT's typical answer names more companies per response than Gemini's does, it would mechanically touch more of the cohort even if both engines were drawing from the same underlying knowledge. Another is retrieval breadth: if one engine's process for generating a buyer-style recommendation pulls from a larger candidate pool before narrowing down to a final answer, it would have more opportunities to surface a long-tail company that a narrower process never considers. A third is simply training data coverage: an engine trained on a less complete picture of a given category would have fewer companies available to mention regardless of how long its answers run.

What this finding does not tell you is whether the companies Gemini does show are better positioned, on average, than the ones ChatGPT shows. The study's position data suggests Gemini actually mentions companies earlier in its answers on average (4.72 versus 5.08 for ChatGPT, see the companion finding on mention position), so a narrower Gemini answer may still place a company more prominently within it. Breadth and prominence are different things, and an engine can be narrow but decisive, or broad but diffuse.

For a company deciding where to spend effort improving its AI search presence, the practical read is this: being absent from Gemini is common enough, at a 32.6% visibility rate across the cohort, that it should not be read as a unique signal of a problem specific to your company. It is closer to the median experience. Being absent from ChatGPT, where visibility runs higher across the cohort at 43.9%, is comparatively more unusual and may be worth investigating on its own. Tracking all three engines separately, rather than one as a proxy for the others, is the only way to tell which situation you are actually in, since no single engine's visibility rate predicts another's for a specific company.