How many brands does AI mention in its answers?
It depends on which one you ask, and when you ask it. Right now, ChatGPT mentions the most brands per response (4.5 brand mentions), and Perplexity mentions the fewest (2.7). The biggest mover is Google AI Mode: it reached 8.8 brand mentions per response in early May, more than any other engine at the time, then dropped to 2.6 within six weeks. ChatGPT and Google AI Overview both started this period unusually high and came down steadily through April. Count each brand only once, and the field gets tighter still: every engine currently names between 2 and 3 different brands per answer, on average.
“Right now” / “current” = the 14 days ending Jul 13, 2026 (Jun 30 – Jul 13). “Since mid-May” / “the stable period” = May 15 – Jul 13, 2026, chosen because that’s when every engine’s numbers had settled into a steadier range. Both defined in full under methodology.
ChatGPT
AI Overview
Perplexity
Gemini
AI ModeHow has this changed since March?
This covers the period since all five engines we track were running at the same time: March 1 through July 13, 2026. We use a 7-day average so daily ups and downs don't hide the bigger pattern. Each line is labeled with its engine name at the end, and the numbered markers are dated announcements from the AI providers, listed under the chart.
What the AI providers changed during this period
Three announcements from the engine providers themselves land inside this window, all within two weeks of each other in May. They’re marked on the chart above. Worth noting: several things changed at once, so timing overlap doesn’t prove any one of them caused a specific move.
OpenAI updated ChatGPT's default model, citing more accurate answers and reduced hallucinations. GPT-5.5 itself had launched about two weeks earlier, on April 23. OpenAI release notes
Google rolled out more links placed directly inside AI responses, website previews, article suggestions at the end of answers, personal perspectives from forums and social, and labels for news subscriptions. All five change how sources appear inside an AI answer. Google blog
Google made Gemini 3.5 Flash the default model in AI Mode globally, and merged AI Overviews and AI Mode into a single flow. This one change affects two of the engines we track. Google's May 2026 core search update also began two days later, on May 21, and ran through June 2. Google blog
What happened to each engine
All values are 7-day averages.
Its numbers were high when this period began, at 8.1 brand mentions per response in early March. It came down through April and has stayed close to 4.2–4.7 since mid-May. Right now, it's the steadiest of the five engines, and the only one mentioning slightly more brands than it did in May.
It's been falling steadily for most of this period, from around 5.3 brand mentions per response in mid-March down to 2.7 today, the fewest of the five.
One of the calmer lines on the chart. It moved down slowly, from 5.3 in early March to a low of 3.1 in mid-June, and has stayed around 3.4–3.5 since. Over the last 8 weeks it's been roughly flat.
This engine took the longest to settle. It opened this period at 10.3 brand mentions per response, the highest value on the chart, and didn't level off until late April. Since then it's stayed in a steadier 3.3–4.5 range, and is currently at 3.5.
This engine moved the most of any we track. It rose from under 6 in March to a peak of 8.8 in early May, then dropped to 2.6 by mid-June. That's the sharpest rise and the sharpest fall of any engine in this period. It's recovered slightly, to 3.2 today.
Which engine mentions the most brands?
ChatGPT does, consistently – both on average since mid-May (4.4 brand mentions per response) and right now (4.5). Perplexity mentions the fewest right now, at 2.7. AI Mode has dropped the most since mid-May: it averaged 3.9 over the period, and is down to 3.2 over just the last two weeks.
ChatGPTGoogle AI OverviewGeminiGoogle AI ModePerplexityHow many different brands is that?
Fewer than the mention counts suggest. Counting each brand only once per response, every engine currently lands between 2.2 and 3.3 different brands per answer. The rest is repetition, and the engines differ in how much they repeat: ChatGPT and Google AI Overview mention each brand about 1.4 times on average right now, while Perplexity is closest to naming a brand once and moving on, at about 1.2. ChatGPT leads either way you count.
ChatGPTGeminiGoogle AI OverviewGoogle AI ModePerplexityAll figures are averages over the last 14 days (Jun 30 – Jul 13, 2026). “Repeats” = brand mentions divided by different brands.
Which engine is the least predictable?
Google AI Mode, by a wide margin. We measured this over the stable period (May 15 – Jul 13, 2026), after ChatGPT and Google AI Overview had settled from their unusually high start to the period, so this reflects real day-to-day noise. AI Mode's swings are about 1.7x bigger than the next most volatile engine (Perplexity, at 18.2%), and more than three times ChatGPT's (9.0%). If you only check your visibility on AI Mode once in a while, a single noisy day can look like a real trend when it isn't.
Google AI ModePerplexityGeminiGoogle AI OverviewChatGPTWhat this means if you're tracking your own brand
Checking the wrong engine on the wrong day can give you a misleading picture. Here's how to read movement on each one.
How did we collect and count this?
Here's exactly what we counted, and how.
For every prompt Vercite runs against an engine, we count every brand mention in the response. If the same brand is mentioned three times in one answer, that counts as three mentions, not one. We average this across all the prompts run for that engine on a given day, which gives us the daily figure used throughout this article. So a higher number can mean an engine is naming more brands, repeating the same brands more often, or both.
In the "How many different brands is that?" section, we use a second count where each brand is counted only once per response, no matter how many times it's mentioned. Both counts come from the same set of responses, so they're directly comparable. Dividing mentions by different brands gives the repeat figure shown in that section.
These prompts come from real Vercite customer data, so they cover a wide mix of languages, industries, and query types rather than a controlled sample. Because of that, we focus on patterns that hold across the whole dataset, like breadth and volatility, rather than breaking results down by industry.
This article is based on just over 234,000 AI responses collected between March 1 and July 13, 2026. Daily volume differs by engine: ChatGPT averages around 530 responses per day, Perplexity around 460, Google AI Overview around 400, Google AI Mode around 240, and Gemini around 110. One thing this rules out: AI Mode's high volatility isn't a small-sample effect. It gets more than twice as many daily responses as Gemini, yet swings far more, so its unpredictability looks like real engine behavior. Gemini's smaller daily sample is worth keeping in mind when reading its day-to-day movements, though.
This piece covers March 1 – July 13, 2026, the period since all five engines we track were running at the same time. "Current" figures use the 14 days ending July 13, 2026 (Jun 30 – Jul 13). When this piece says "the stable period" or "since mid-May," we mean May 15 – July 13, 2026 specifically, once every engine's numbers had settled into a steadier range.
Try Vercite for free today
Monitor your brand across ChatGPT, Gemini, Perplexity & Google AI