ChatGPT search research · Comparison

Interface vs API

ChatGPT Plus in the app against the OpenAI API with web search: what the API shows and what people see.

Wave 2026-11Plus: gpt-6-thinkingAPI: gpt-6-sol100 prompts on bothNew York, US
Answer brands the first search did not name
47%Plus
53%API
They came after the first search, so the web had a chance to put them there.
Search steps per answer
1.2Plus
2.6API
API searches 2.1× as much, and more on 82 of 100 prompts.
First search already names a brand
78%Plus
81%API
78 of 100 against 81 of 100 answers: candidates chosen before searching.
Answers that opened a page
15%Plus
39%API
91% of the pages the API cited were never opened, only seen in results.
Brands they agree on
39%of brands, same prompt
For the same prompt: 243 brands named by both, 278 by Plus only, 147 by API only.
  • API runs about 2.1 times as many search steps per answer (1.2 for Plus, 2.6 for API).
  • The first search already names a brand in 78 of 100 Plus answers and 81 of 100 API answers.
  • Of the brands in the answer, 47% on Plus and 53% on API were not in its first search, so search results had a chance to put them there. The rest were in its first search, named before any results.
  • "Best …" articles make up 33% of Plus's citations and 9% of API's; brands' own sites 36% and 60%.
  • The API opened a page in 39 of 100 answers; most of what it cites it saw only as a search result.
  • For the same prompt they agree on 39% of the brands they name between them.
PlusAPI

Every prompt ends by asking ChatGPT to list the searches it ran. The app hides the text of some searches, mostly the second-layer fan-outs it runs while reasoning, so for the app's query numbers this self-report is the default: it is used when it contains every query seen in the stream (Plus 100 of 100 answers), otherwise only what was seen counts. Where the self-report changes a number, the number without it is shown under it. Search steps, page reads, citations and brands always come from the stream.

How much it searches

Counted from search events, which both surfaces show even when the app hides the query text.

Seen: events in the app's stream or calls in the API response.

Search steps per answer
A step is one search call; it can send several queries at once.
Same prompt: Plus higher 1, API 82, equal 17
Plus 1.2median 1
API 2.6median 3
Answers with two or more steps
ChatGPT searched, read the results and searched again.
Plus 24%24 of 100
API 87%87 of 100
Answers that searched at all
Plus 100%100 of 100
API 100%100 of 100
Steps with the query text hidden
The step is in the stream, its query text is not.
Plus 19%24 of 124 steps
API 0%0 of 263 steps

The first search

Does ChatGPT start from brands it already has in mind, or from an open search? The first search step decides it.

A step can hold several queries. It counts as naming a brand when at least one of them does. The step is taken from the stream on every surface. Round 1 of the app's self-report fills in only a first step whose text was hidden: that round also lists product lookups by name that ran in no visible step, so it would overstate how often the first search names a brand.

In this wave every first search showed its query text. When the app hides it, the answer is counted as unknown here rather than judged by its second search.

Names a brand or product
At least one query in the first step names one: the candidates were chosen before searching.
Plus 78%78 of 100
API 81%81 of 100
Names no brand
A "best …" or topic search, open to whatever comes back.
Plus 22%22 of 100
API 19%19 of 100
First-step queries that name a brand
Share of the queries in the first step, not of answers.
Plus 63%191 of 305 queries
API 56%200 of 358 queries

From memory to the answer

The first search runs before ChatGPT has seen any results, so the brands it names there come from what the model already knows. How many of them survive into the answer, and how much of the answer they make up?

Brands are matched loosely: “RAV4” in a query matches “Toyota RAV4 Hybrid” in the answer. Answers that did not search, or whose first search hid its text, are left out. On the app, a hidden first step is filled from round 1 of the self-report.

Brands named in the first search
Per answer whose first search is known. Before any results, so they come from what the model already knows.
Plus 3.0median 3
API 2.4median 3
First-search brands that make the answer
Of the brands it searched for first, the share it names in the answer.
Plus 85%254 of 300 brands
API 68%165 of 244 brands
Answer brands that were in the first search
Of the brands in the answer, the share it already had in mind before searching. The rest came later: from results or later searches.
Plus 53%275 of 521 brands
API 47%183 of 390 brands
Answers that drop a first-search brand
Among answers whose first search named a brand: at least one of those brands is missing from the answer.
Plus 44%34 of 78
API 67%54 of 81

Search journeys

A journey is the searches of one answer in order. Two questions sort them: does the first search already name brands, and does ChatGPT search again once it has seen results?

Most common: names brands, searches once on Plus (57% of answers), names brands, then searches again on API (72%).

A second search is counted from the search steps in the stream, or from a later round of the self-report on the app. Brands are matched loosely, as above.

How the journeys go
Share of answers; the rows add up to 100%.
Names brands, searches once
The first search already names the brands and there is no second one: it checks its own picks in one go, then writes.
Plus 57%57 of 100
API 9%9 of 100
Names brands, then searches again
The first search names brands; after the results come back it searches again.
Plus 21%21 of 100
API 72%72 of 100
Starts open, names brands later
The first search names no brand. Brands appear only in a later search, after it has seen results.
Plus 2%2 of 100 0% · 0 of 100 without the self-report
API 10%10 of 100
Never searches for a brand
No search names a brand: it takes the options from the lists and pages it found.
Plus 20%20 of 100 22% · 22 of 100 without the self-report
API 9%9 of 100
What the searches after the first one do
Among answers that searched again and showed what they typed (Plus 22, API 87). One answer can do several of these.
Names a brand the first search did not
Picked up from the results it had just seen, or recalled later.
Plus 91%20 of 22
API 77%67 of 87
Searches a first-search brand again
It goes back to one of its own picks to check it.
Plus 82%18 of 22
API 71%62 of 87
Looks up facts or prices
Plus 73%16 of 22
API 72%63 of 87
Checks a brand on its own
Plus 77%17 of 22
API 76%66 of 87
Checks rules or official sources
Plus 18%4 of 22
API 55%48 of 87
Asks for reviews or forums
Plus 5%1 of 22
API 18%16 of 87
Searches for a list or the topic again
Plus 18%4 of 22
API 7%6 of 87
One prompt, two journeys: “best online car selling sites”
A prompt where each surface took its most common journey. Up to 3 queries per search are shown; the answer page has all of them.
Plus names brands, searches once, as ChatGPT listed its searches
  1. 3 queries · List, Facts, Compare
    • best websites sell your car online 2026 CarMax Carvana Kelley Blue Book Edmunds private sale
    • Carvana CarMax sell my car online offer pickup 2026 official
    • Autotrader Cars.com Facebook Marketplace private party sell car listing fees 2026
Every query of this answer →
API names brands, then searches again
  1. 4 queries · Brand check, Facts
    • site:carmax.com sell my car online offer valid 7 days appointment
    • site:carvana.com sell my car online offer pickup payment
    • site:cars.com sell my car private listing fees 2026
    • and 1 more
  2. 4 queries · Brand check, Facts, Reviews
    • site:cars.com sell my car private listing free sell to dealer offer
    • site:cargurus.com sell my car instant offer private listing fees
    • site:facebook.com/help marketplace sell vehicle listing fee
    • and 1 more
  3. 3 queries · Facts
    • site:facebook.com/help sell vehicle Facebook Marketplace local pickup listing fees vehicles
    • site:bringatrailer.com how it works sellers listing fees 2026
    • site:kbb.com instant cash offer sell car participating dealer 7 days
Every query of this answer →

Query kinds

The mix of its queries: each slice is the share of its queries of that kind. Each prompt's answer page shows every query with its kind.

Most common: brand check on Plus, fact lookup on API.

Kinds come from the classifier's query labels. On the app the mix includes the self-report, which also covers the steps whose text the app hid; the table gives the mix without it. Lookups the self-report puts in its first round but the stream shows in no step count under all searches only.

Every query of every answer.

Plus
Topic search: 1% (5 of 537 queries)Open list search: 10% (53 of 537 queries)List search with its own picks: 12% (63 of 537 queries)Brand check: 36% (194 of 537 queries)Fact lookup: 26% (138 of 537 queries)Comparison: 2% (12 of 537 queries)Reviews and forums: 3% (18 of 537 queries)Other: news, official rules, not labeled: 10% (54 of 537 queries) 36% Brandcheck 26% Factlookup 12% List withbrands 537 queries Topic search: 1% (5 of 537 queries)Open list search: 10% (53 of 537 queries)List search with its own picks: 12% (63 of 537 queries)Brand check: 36% (194 of 537 queries)Fact lookup: 26% (138 of 537 queries)Comparison: 2% (12 of 537 queries)Reviews and forums: 3% (18 of 537 queries)Other: news, official rules, not labeled: 10% (54 of 537 queries) 36% Brand check 26% Fact lookup 12% List withbrands 537 queries
API
Topic search: 0% (1 of 1010 queries)Open list search: 5% (48 of 1010 queries)List search with its own picks: 3% (26 of 1010 queries)Brand check: 28% (281 of 1010 queries)Fact lookup: 40% (403 of 1010 queries)Comparison: 1% (13 of 1010 queries)Reviews and forums: 5% (46 of 1010 queries)Other: news, official rules, not labeled: 19% (192 of 1010 queries) 40% Factlookup 28% Brandcheck 19% Other 1010 queries Topic search: 0% (1 of 1010 queries)Open list search: 5% (48 of 1010 queries)List search with its own picks: 3% (26 of 1010 queries)Brand check: 28% (281 of 1010 queries)Fact lookup: 40% (403 of 1010 queries)Comparison: 1% (13 of 1010 queries)Reviews and forums: 5% (46 of 1010 queries)Other: news, official rules, not labeled: 19% (192 of 1010 queries) 40% Fact lookup 28% Brand check 19% Other 1010 queries
Plus
Topic search: 1% (5 of 537 queries)Open list search: 10% (53 of 537 queries)List search with its own picks: 12% (63 of 537 queries)Brand check: 36% (194 of 537 queries)Fact lookup: 26% (138 of 537 queries)Comparison: 2% (12 of 537 queries)Reviews and forums: 3% (18 of 537 queries)Other: news, official rules, not labeled: 10% (54 of 537 queries) 537 queries
API
Topic search: 0% (1 of 1010 queries)Open list search: 5% (48 of 1010 queries)List search with its own picks: 3% (26 of 1010 queries)Brand check: 28% (281 of 1010 queries)Fact lookup: 40% (403 of 1010 queries)Comparison: 1% (13 of 1010 queries)Reviews and forums: 5% (46 of 1010 queries)Other: news, official rules, not labeled: 19% (192 of 1010 queries) 1010 queries
Kind Plus API
Topic search 1% 5 1% without the self-report 0% 1
Open list search 10% 53 14% without the self-report 5% 48
List search with its own picks 12% 63 15% without the self-report 3% 26
Brand check 36% 194 11% without the self-report 28% 281
Fact lookup 26% 138 34% without the self-report 40% 403
Comparison 2% 12 4% without the self-report 1% 13
Reviews and forums 3% 18 6% without the self-report 5% 46
Other: news, official rules, not labeled 10% 54 16% without the self-report 19% 192

Only the first search step of each answer, before it had seen any results.

Plus
Topic search: 1% (2 of 305 queries)Open list search: 14% (42 of 305 queries)List search with its own picks: 15% (47 of 305 queries)Brand check: 11% (33 of 305 queries)Fact lookup: 34% (104 of 305 queries)Comparison: 4% (12 of 305 queries)Reviews and forums: 6% (17 of 305 queries)Other: news, official rules, not labeled: 16% (48 of 305 queries) 34% Factlookup 16% Other 15% List withbrands 305 queries Topic search: 1% (2 of 305 queries)Open list search: 14% (42 of 305 queries)List search with its own picks: 15% (47 of 305 queries)Brand check: 11% (33 of 305 queries)Fact lookup: 34% (104 of 305 queries)Comparison: 4% (12 of 305 queries)Reviews and forums: 6% (17 of 305 queries)Other: news, official rules, not labeled: 16% (48 of 305 queries) 34% Fact lookup 16% Other 15% List withbrands 305 queries
API
Topic search: 0% (1 of 358 queries)Open list search: 12% (42 of 358 queries)List search with its own picks: 6% (23 of 358 queries)Brand check: 15% (52 of 358 queries)Fact lookup: 38% (136 of 358 queries)Comparison: 3% (9 of 358 queries)Reviews and forums: 6% (20 of 358 queries)Other: news, official rules, not labeled: 21% (75 of 358 queries) 38% Factlookup 21% Other 15% Brandcheck 358 queries Topic search: 0% (1 of 358 queries)Open list search: 12% (42 of 358 queries)List search with its own picks: 6% (23 of 358 queries)Brand check: 15% (52 of 358 queries)Fact lookup: 38% (136 of 358 queries)Comparison: 3% (9 of 358 queries)Reviews and forums: 6% (20 of 358 queries)Other: news, official rules, not labeled: 21% (75 of 358 queries) 38% Fact lookup 21% Other 15% Brand check 358 queries
Plus
Topic search: 1% (2 of 305 queries)Open list search: 14% (42 of 305 queries)List search with its own picks: 15% (47 of 305 queries)Brand check: 11% (33 of 305 queries)Fact lookup: 34% (104 of 305 queries)Comparison: 4% (12 of 305 queries)Reviews and forums: 6% (17 of 305 queries)Other: news, official rules, not labeled: 16% (48 of 305 queries) 305 queries
API
Topic search: 0% (1 of 358 queries)Open list search: 12% (42 of 358 queries)List search with its own picks: 6% (23 of 358 queries)Brand check: 15% (52 of 358 queries)Fact lookup: 38% (136 of 358 queries)Comparison: 3% (9 of 358 queries)Reviews and forums: 6% (20 of 358 queries)Other: news, official rules, not labeled: 21% (75 of 358 queries) 358 queries
Kind Plus API
Topic search 1% 2 1% without the self-report 0% 1
Open list search 14% 42 14% without the self-report 12% 42
List search with its own picks 15% 47 15% without the self-report 6% 23
Brand check 11% 33 11% without the self-report 15% 52
Fact lookup 34% 104 34% without the self-report 38% 136
Comparison 4% 12 4% without the self-report 3% 9
Reviews and forums 6% 17 6% without the self-report 6% 20
Other: news, official rules, not labeled 16% 48 16% without the self-report 21% 75

Only the searches after the first, once it had seen results. Answers that searched once have none.

Plus
List search with its own picks: 4% (5 of 113 queries)Brand check: 60% (68 of 113 queries)Fact lookup: 29% (33 of 113 queries)Reviews and forums: 1% (1 of 113 queries)Other: news, official rules, not labeled: 5% (6 of 113 queries) 60% Brandcheck 29% Factlookup 5% Other 113 queries List search with its own picks: 4% (5 of 113 queries)Brand check: 60% (68 of 113 queries)Fact lookup: 29% (33 of 113 queries)Reviews and forums: 1% (1 of 113 queries)Other: news, official rules, not labeled: 5% (6 of 113 queries) 60% Brand check 29% Fact lookup 5% Other 113 queries
API
Open list search: 1% (6 of 652 queries)List search with its own picks: 0% (3 of 652 queries)Brand check: 35% (229 of 652 queries)Fact lookup: 41% (267 of 652 queries)Comparison: 1% (4 of 652 queries)Reviews and forums: 4% (26 of 652 queries)Other: news, official rules, not labeled: 18% (117 of 652 queries) 41% Factlookup 35% Brandcheck 18% Other 652 queries Open list search: 1% (6 of 652 queries)List search with its own picks: 0% (3 of 652 queries)Brand check: 35% (229 of 652 queries)Fact lookup: 41% (267 of 652 queries)Comparison: 1% (4 of 652 queries)Reviews and forums: 4% (26 of 652 queries)Other: news, official rules, not labeled: 18% (117 of 652 queries) 41% Fact lookup 35% Brand check 18% Other 652 queries
Plus
List search with its own picks: 4% (5 of 113 queries)Brand check: 60% (68 of 113 queries)Fact lookup: 29% (33 of 113 queries)Reviews and forums: 1% (1 of 113 queries)Other: news, official rules, not labeled: 5% (6 of 113 queries) 113 queries
API
Open list search: 1% (6 of 652 queries)List search with its own picks: 0% (3 of 652 queries)Brand check: 35% (229 of 652 queries)Fact lookup: 41% (267 of 652 queries)Comparison: 1% (4 of 652 queries)Reviews and forums: 4% (26 of 652 queries)Other: news, official rules, not labeled: 18% (117 of 652 queries) 652 queries
Kind Plus API
Open list search 0% 0 – without the self-report 1% 6
List search with its own picks 4% 5 – without the self-report 0% 3
Brand check 60% 68 – without the self-report 35% 229
Fact lookup 29% 33 – without the self-report 41% 267
Comparison 0% 0 – without the self-report 1% 4
Reviews and forums 1% 1 – without the self-report 4% 26
Other: news, official rules, not labeled 5% 6 – without the self-report 18% 117

What it searches for

Share of answers with at least one query of each type.

“Words” rows are read from the text of the query, so anyone can check them. “Classifier” rows come from the query labels. On the app the numbers include the self-report; the number without it follows where it differs.

Queries per answer
Without the self-report the app count is a floor: hidden steps have no text.
Plus 5.4median 5 3.1 · median 3 without the self-report
API 10.1median 11
Names a brand or productclassifier
At least one query names a brand or product (publications such as NerdWallet do not count).
“TuxMat official custom car mats coverage clips waterproof cleaning”
Plus 92%92 of 100 78% · 78 of 100 without the self-report
API 91%91 of 100
Asks about price, fees or rateswords
The query contains a word such as price, cost, MSRP, fee, APY, APR, rate or a $ amount.
“site:omniluxled.com/products/omnilux-contour-face price 633 830”
Plus 56%56 of 100 53% · 53 of 100 without the self-report
API 66%66 of 100
Asks for reviews, Reddit or forumswords
The query contains review, Reddit, forum, Quora, Trustpilot or complaints.
“best 2027 4WD SUVs off road 2026 reviews Land Cruiser 4Runner Wrangler Bronco”
Plus 27%27 of 100
API 24%24 of 100
Limits a search to one site (site:)words
The query uses the site: operator.
“site:shopify.com blog reorder point formula safety stock inventory”
Plus 17%17 of 100 6% · 6 of 100 without the self-report
API 96%96 of 100
Searches inside a brand's own siteclassifier
site: pointing at a domain labeled as a brand's own site.
“site:belkins.io lead generation services small businesses pricing”
Plus 16%16 of 100 5% · 5 of 100 without the self-report
API 84%84 of 100

Pages it reads

Opening a page means reading it beyond the title and snippet in the search results.

The last two rows read the overlap both ways: how much of what it cites it had read, and how much of what it read it cites.

Answers that opened a page
The API's open_page or find_in_page call; a "view" result in the app stream.
Plus 15%15 of 100
API 39%39 of 100
Pages opened per answer
Plus 0.5median 0
API 0.5median 0
Cited pages it had opened
The rest were cited from the search results alone: title and snippet.
Plus 6%40 of 624 cited
API 9%34 of 398 cited
Opened pages it went on to cite
The rest were read but left out of the citations.
Plus 87%40 of 46 opened
API 65%34 of 52 opened

What reaches the answer

What the reader gets: citations, links in the text, the sites behind them and the brands named.

For the same prompt, the two share 26% of their cited sites on average.

Cited pages per answer
Same prompt: Plus higher 71, API 20, equal 9
Plus 6.2median 6
API 4.0median 4
Cited sites per answer
Different domains among the cited pages.
Plus 4.9median 5
API 3.4median 3
Brands and products named
Same prompt: Plus higher 71, API 10, equal 19
Plus 5.2median 5
API 3.9median 4
Answer length, characters
Includes the list of searches each prompt asks for.
Plus 6,474
API 1,852
Links in the answer text
Links ChatGPT writes into the text, apart from its citations. In the API the citations are themselves the links in the text, so its links here are only bare URLs outside them.
Answers with a link in the text
Plus 45%45 of 100
API 0%0 of 100
Linked pages per answer
Same prompt: Plus higher 45, API 0, equal 55
Plus 2.1median 0
API 0.0median 0
Linked pages on sites it did not cite
Linked in the text, and no page of that site is among the answer's citations. Mostly the homepages of brands it names.
Plus 56%120 of 213 linked
API –
Linked pages that are homepages
A link to the front page of a site rather than to a page about the question.
Plus 52%110 of 213 linked
API –
Linked pages never seen in search results
ChatGPT linked a page that did not come back in any of its searches.
Plus 61%130 of 213 linked
API –
Plus: 573 cited only, 51 cited and linked, 162 linked onlyAPI: 398 cited only, 0 cited and linked, 0 linked only
The kind of sites cited
Share of each surface's cited pages, by the type of site.
Brand sites
Plus 36%226 of 624
API 60%237 of 398
Reference and government
Plus 21%128 of 624
API 21%84 of 398
Editorial
Plus 30%185 of 624
API 7%28 of 398
Marketplaces
Plus 5%34 of 624
API 9%35 of 398
Reviews
Plus 3%21 of 624
API 2%9 of 398
News
Plus 4%24 of 624
API 0%1 of 398
Community (UGC)
Plus 0%3 of 624
API 1%3 of 398
Other
Plus 0%3 of 624
API 0%1 of 398
Big platforms: in the results, and cited
How much of what its searches return comes from the big platforms, and how much of that it cites. Each platform's row is its share of the search results, with its results and citations beside it.
Search results from big platforms
Pages from Reddit, Wikipedia, YouTube, Amazon, Quora, Facebook, Instagram, TikTok, LinkedIn, X, Pinterest, Medium, Yelp, Tripadvisor, among all the pages its searches returned.
Plus 4%218 of 5137 results
API 17%928 of 5308 results
Cited pages from big platforms
The same sites among the pages it cited.
Plus 0%3 of 624 cited
API 1%2 of 398 cited
Reddit in the results
Plus 3%166 results · 1 cited
API 14%724 results · 0 cited
Wikipedia in the results
Plus 0%10 results · 0 cited
API 3%167 results · 0 cited
YouTube in the results
Plus 0%21 results · 0 cited
API 1%27 results · 0 cited
Amazon in the results
Plus 0%21 results · 2 cited
API 0%6 results · 2 cited
Facebook in the results
Plus 0%0 results · 0 cited
API 0%2 results · 0 cited
Yelp in the results
Plus 0%0 results · 0 cited
API 0%2 results · 0 cited
"Best …" articles: in the results, and cited
How much of what its searches return are list articles, and how much of what it cites. Then who publishes them: each group's share of the list articles in the results, with its results and citations beside it.
Search results that are "best …" articles
List articles among all the pages its searches returned.
Plus 25%1262 of 5137 results
API 11%593 of 5308 results
Cited pages that are "best …" articles
List articles among the pages it cited.
Plus 33%206 of 624 cited
API 9%36 of 398 cited
Published by media and review sites
Editorial, review and news sites.
Plus 87%1101 results · 180 cited
API 89%528 results · 27 cited
Published by brands on their own sites
A brand's own "best …" list, usually with the brand on it.
Plus 7%89 results · 9 cited
API 7%43 results · 5 cited
Published elsewhere
Marketplaces, community and reference sites, and sites not yet labeled.
Plus 6%72 results · 17 cited
API 4%22 results · 4 cited
Brands named for the same prompt
668 brands across 100 prompts; on average they agree on 39% per prompt.
Plus only: 278Both: 243API only: 147

What the web can move

How much of the answer is open to what search finds, rather than fixed by what the model already knows. Content cannot change the model's memory, but it can change what these searches return.

The rows below are the ways in, each a share that can be checked on the answer pages. Page types and site types come from the classifier's labels. Cited pages that are among the search results the app shows: Plus 622 of 624.

Answer brands the first search did not name
47% Plus 246 of 521 brands
53% API 207 of 390 brands
The first search runs before any results, so a brand it did not name came later: the web had a chance to put it there, though some may still come from memory. Answers whose first search is known.
Where the web gets in
Each row is one opening for content, with what a brand would do about it.
Answers with an open list search
A "best …" search that names no brand: the pages it returns supply the candidates. Ranking on them is the way in.
Plus 35%35 of 100 34% · 34 of 100 without the self-report
API 34%34 of 100
Cited pages that are "best …" articles
List articles it cites. Being on these lists puts a brand in front of it.
Plus 33%206 of 624 cited
API 9%36 of 398 cited
Answer brands that no search named
They came from the pages it read, not from its queries. Answers that searched.
Plus 37%187 of 507 brands 48% · 244 of 507 brands without the self-report
API 35%130 of 376 brands
Cited pages from brands' own sites
Product, pricing and help pages that answer its fact lookups and brand checks. A brand controls these pages.
Plus 36%226 of 624 cited
API 60%237 of 398 cited
Searches that added no cited page
No page it cited first came back in that search step: it looked and used nothing new. A page that answers such a search well meets little competition.
Plus 5%6 of 124 steps
API 22%58 of 263 steps

Every prompt in detail

Each prompt with its searches, citations, links and brands on all three surfaces is on its category page.

Read with care

What these numbers can and cannot say.

  • The prompts decide most of what these numbers can say. They are 100 commercial searches such as “best credit cards for travel” or “where to open a roth ira”, typed word for word as US searchers type them, in five categories. There are no comparison, advice or opinion prompts. Read every share as true for prompts like these, not for ChatGPT in general.
  • One run per prompt and surface. ChatGPT does not answer the same way twice, so a single prompt can differ by chance; averages over many prompts are what to read.
  • The surfaces run different models (Plus: gpt-6-thinking; API: gpt-6-sol, medium effort). A difference between them is the surface and the model together.
  • The app hides the query text of some search steps (Plus 19%). Query numbers for the app include ChatGPT's own list of its searches to fill that gap; a list the model wrote can miss or add a search.
  • Search results are not compared: the API returns every result it got, the app only what it shows.