Cars have the most agreement, mobile carriers the least.
All three assistants share a sole first-name leader in 15 of 20 countries for cars, but only 3 for mobile carriers, the lowest agreement of the nine categories.
October 2026 AIGROW Index
The brands AI recommends in 20 countries
We asked ChatGPT, Gemini and Claude about banks, cars, supermarkets and six other consumer categories. Across 5,040 answers, the first brand changes with the assistant, the question and the country.
Study by Guglielmo Paesano, founder of AIGROW · data under CC BY 4.0
Countries with the same sole first-name leader across all three assistants
- Car brands
- 15 / 20
- Online brokers
- 10 / 20
- Electronics retailers
- 10 / 20
Four questions combined, without web search. Ties do not count as agreement.
The first choice changes at the border
Choose a category to see where ChatGPT, Gemini and Claude share a first-name leader. Ties stay visible. This measures answer ordering, not which brand is best.
Same sole leader in 6 of 20 countries
Each tile is one sampled country. Focus a country to see its leaders. Open its link for all nine categories. Positions are approximate, not to scale.
=Same sole leader≠Different leaders?Tie or no leader
Italy
- ChatGPT
- FinecoBank6 of 12 first names
- Gemini
- Intesa Sanpaolo6 of 12 first names
- Claude
- FinecoBank2 of 4 first names
Compare all 20 countries in this category
Four questions combined. ChatGPT and Gemini have twelve answers each per country; Claude has four. Equal first-name counts are ties, and do not count as agreement.
Four findings worth quoting
Every figure comes from this frozen sample. Pooled answer counts give ChatGPT and Gemini three times the weight of Claude.
Toyota leads across all 20 countries.
Toyota is the sole first-name leader for at least one assistant in every country. It comes first in 278 of 560 car answers, compared with 104 for Tesla.
Energy has the most three-way splits.
In 6 of 20 countries, each assistant has a different sole first-name leader for energy suppliers, more than in any other category.
In Italy, Esselunga comes before Lidl.
Esselunga comes first in 21 of 28 supermarket answers about Italy. Lidl comes first in 0. This is an ordering count, not a measure of store quality.
Read the Italy resultsPress kit
Use the country URL for local results and the DOI for the frozen October Index dataset. The study files below are derived from that edition; they do not have a separate DOI.
- Successful answers
- 5,040
- Countries
- 20
- Categories
- 9
Guglielmo Paesano (2026). The brands AI recommends in 20 countries. AIGROW. October 2026 Index data. https://aigrow.gg/research/ai-brands-2026 DOI 10.5281/zenodo.23188442
Download a country chart
Preview the chart
What we measured
A frozen sample of consumer-brand answers from the October AIGROW Index, not a survey of buyers and not an evaluation of the brands.
Collected from 2026-10-05 to 2026-10-06: 5,040 successful answers across 20 countries and 9 categories. Each cell combines four locally worded questions. ChatGPT and Gemini answered each question three times; Claude once. This study excludes web-search answers and Grok.
Collection, counting and model details
The writer reads the already published October pages and the latest stored answer for each probe up to 2026-10-06T23:59:59.999Z. It checks every published brand count against that page before writing. No new questions, model calls or Index aggregation are run.
A brand counts once per answer, using the Index spellings and stored aliases. First means the first distinct name in the extracted list. Names are not newly merged across countries; a local subsidiary or differently named brand may remain separate. Unlike the public rankings, the downloadable study retains names seen only once.
There were 27 successful answers with no brand. They remain in the denominators. Failed calls are excluded. A tie is reported as a tie, and counts are descriptive: repeated answers are not independent consumer observations.
- ChatGPT
- gpt-6-sol
- Gemini
- google/gemini-3.1-flash-lite
- Claude
- claude-sonnet-5
ChatGPT and Claude were collected through subscription command-line tools; Gemini through an API gateway. The country was named in the local-language prompt. We did not test geolocated public accounts. Model updates, personalization, the wording of a question and web search can change the answer.
Read the full Index methodologyQuestions about the study
Does first place mean the best brand?
No. First means the first brand in the extracted list of an answer. It measures ordering in this sample, not product quality, suitability, market share or a promise of a future recommendation.
Why do the assistants have different answer counts?
Each country and category has four questions. ChatGPT and Gemini answered each three times, giving twelve answers each. Claude answered once, giving four. Pooled totals therefore give ChatGPT and Gemini more weight; compare assistants separately in the country tables.
Can I reuse the charts and data?
Yes, under CC BY 4.0. Credit AIGROW, link to this study and include the October dataset DOI. Each country page offers a chart and an image embed code with a link back. The CSV includes every counted brand, including brands named only once.
Do these answers represent the public chatbot apps today?
They are a frozen October 2026 sample from specific models and collection routes, without web search. Answers in a public app can differ with its model, location, personalization, prompt and date. We did not survey consumers or measure purchases.
Check what AI can read about your business.
The free scan checks your website for signals that help assistants understand it. Your results are separate from this consumer-brand study.