Who wins the 2026 World Cup according to ChatGPT? To answer it without opinions, in CreceRank we ran the same prompt 400 times and counted which team it chose as champion in each response. What we really wanted to know was whether the AI always repeats the same thing or if it changes a lot with each query. The result: ChatGPT leans towards France (31% of the time), although its response is less firm, and quite more curious, than it seems.
Asking ChatGPT once who wins the 2026 World Cup isn't very useful: language models are probabilistic and each answer can change. At CreceRank, that variability is our daily work, because we measure how brands appear (or not) within the AI's responses. So we asked the underlying question: does ChatGPT always give the same answer or does it change a lot from one query to another? To find out, we didn't ask once, we asked 400 times. And to see if location influences, we distributed those queries across four countries of origin: Argentina, Mexico, the United States, and Uruguay.
It's not the first time we've tested the AI this way. We already did it when analyzing what sources the AI uses for its recommendations in Spanish. This time the guinea pig was football.
The prompt was always exactly the same, designed to force a single answer without ambiguities:
"Act as a data sports analyst. Analyze the 2026 World Cup as of 06/21/26 and the history of each team. Without beating around the bush, lists, or ambiguities, choose a single champion country. Justify your verdict weighing 60% historical data and 40% current performance in the tournament."
At CreceRank, we live to measure whether AI is consistent. If ChatGPT always answered the same, it would be enough to ask once. Since it changes, we need to repeat and count. The World Cup was the perfect excuse to show how much a single question can vary.
ChatGPT does not "predict" the World Cup: it reflects a historical and statistical consensus that, repeated 400 times, points to France as the favorite, but distributes the rest among the usual powers and never favors the country from which it is asked.
Adding the 400 queries, France is the team that ChatGPT chooses as champion most frequently: 124 times, 31% of the total. Behind are Germany, Argentina, Brazil, and Spain, in a fairly close pack.
| Team | Times chosen (out of 400) | % of queries |
|---|---|---|
| 🇫🇷 France | 124 | 31% |
| 🇩🇪 Germany | 83 | 21% |
| 🇦🇷 Argentina | 76 | 19% |
| 🇧🇷 Brazil | 64 | 16% |
| 🇪🇸 Spain | 53 | 13% |
Champion of the 2026 World Cup chosen by ChatGPT in 400 identical queries (06/21/2026).
Here is the most interesting part of the experiment, and what we really wanted to measure. The same question, asked from four countries, does not always return the same champion. France wins from Argentina, Mexico, and the United States, but from Uruguay, ChatGPT crowns Germany.
| Team | From AR | From MX | From US | From UY |
|---|---|---|---|---|
| 🇫🇷 France | 38% | 31% | 35% | 20% |
| 🇩🇪 Germany | 16% | 25% | 10% | 32% |
| 🇦🇷 Argentina | 15% | 18% | 20% | 23% |
| 🇧🇷 Brazil | 20% | 11% | 18% | 15% |
| 🇪🇸 Spain | 11% | 15% | 17% | 10% |
% of times each team was chosen as champion, according to the country of origin of the query. In green, the winner of each column.
The difference between origins is statistically significant (chi-square test: χ²=26.8; df=12; p=0.008). That is: the place from where you ask does move the needle. However, it is advisable to read it carefully. The four batches were run at different times of the same day, and since ChatGPT uses live web search, part of that variation may be due to the timing of the query and not just to the country. The volatility of Germany, which goes from 10% in the U.S. to 32% in Uruguay, points exactly in that direction.
We would expect that asking from Mexico would give a slight boost to Mexico, or that from Uruguay the Celeste would appear. The exact opposite happened: ChatGPT never chose the local host country. Zero times, in three out of four cases.
Not even Argentina, which is a real contender and current champion, received a local boost: only 15% from Argentina. In fact, it received more votes from Uruguay (23%) than from its own country. ChatGPT does not favor the host: it prioritizes historical and sporting weight over any indication of location.
If your plan was to headline "ChatGPT says my country wins the World Cup", the data contradicts it. The real story is stronger: the AI ignores locality and, repeated hundreds of times, returns a historical consensus that rewards European powers and Brazil.
The key lies in the prompt itself: it asks to weigh 60% historical data and 40% current performance. That 60% historical is what tips the balance towards teams with more finals, titles, and consistency in modern World Cups.
Let's be honest, because it matters: this is not a sports prediction, it's an experiment on how AI responds. Three nuances that are worth clarifying before taking it literally:
The value of the study is not to know who wins, but to understand how AI behaves when consulted en masse: how much it varies, what biases it has and what it does not. It is exactly the type of measurement we do at CreceRank for brand visibility in AI, just like when we study what sources AI cites in Spanish.
Each of the 400 runs was recorded in CreceRank with its complete response. Here are some, just as ChatGPT returned them:
What matters about the exercise is the method: repeat, count, and measure instead of settling for a single answer. It is exactly the logic with which we at CreceRank measure brand visibility within AI, and the reason we set up the entire experiment in our own tool. If you're interested in the topic, we also break down what sources the AI uses in Spanish in our first study.
In an experiment by CreceRank with 400 identical queries, ChatGPT chose France as the champion of the 2026 World Cup 31% of the time, ahead of Germany (21%), Argentina (19%), Brazil (16%), and Spain (13%). France is its most frequent answer, although not a majority: 69% of the time it chose another team.
Not as a favorite. Argentina was chosen as champion in 19% of the 400 queries, in third place behind France and Germany. Interestingly, it received fewer selections when asked from Argentina (15%) than from Uruguay (23%): ChatGPT does not favor the country from which it is asked.
Yes. France won when asked from Argentina, Mexico, and the United States, but from Uruguay ChatGPT chose Germany. The difference is statistically significant (χ²=26.8; p=0.008), although it is partially linked to the timing of each query, as ChatGPT uses live web search.
No. That was the motivation for the experiment. By repeating the same prompt 400 times, ChatGPT chose five different selections as champions and changed favorites based on the origin and timing of the query. Therefore, a single answer is not representative: you need to ask many times and measure the distribution, which is what CreceRank does.
It is not advisable to take it as a sports forecast. It is an experiment on how the AI responds: the favorite appears only 31% of the time, positions 2 to 5 are tied within the margin of error, and the result changes depending on the day because it relies on web search. It serves to understand the biases of the AI, not for betting.
The same one was always used: "Act as a sports data analyst. Analyze the World Cup 2026 as of 06/21/26 and the history of each team. Without beating around the bush, lists, or ambiguities, choose a single champion country. Justify your verdict weighing 60% historical data and 40% current performance in the tournament." It was executed 100 times per country, in four countries of origin, from CreceRank.
Because the prompt weighs 60% historical data and 40% current performance. This historical weight favors France, which combines two recent finals (champion 2018, finalist 2022) with a deep squad, the profile that scores best in that model. Brazil and Germany enter for their records, and Argentina is penalized against the Europeans despite being the reigning champion.
What we did with the World Cup (asking hundreds of times and measuring the result) is done by CreceRank with your brand: it tracks how you appear in ChatGPT, Perplexity, Google AI Overview and Gemini, with data from Spain and LATAM.
No credit card required · Prompts in Spanish · Data from Spain and LATAM
