#广场预测世界杯赢40000U AI predicts the World Cup: Algorithms tell you who has the best chance of winning
The 2026 North America, Mexico, and Canada World Cup is in full swing, with 48 teams battling it out across the continent.
On the field, players compete with skill and willpower; off the field, an even more "hardcore" competition has already begun—various AI large models are stepping into the arena, becoming "football experts."
From Goldman Sachs' complex economic models on Wall Street, to Opta's supercomputers performing trillions of calculations per second, and to domestic and international AI models like ChatGPT, DeepSeek, Kimi, Doubao, Qianwen, and others—each is predicting the final winner of this World Cup in its own way.
Who does AI favor? Are there disagreements among the models? Are their predictions reliable?
Goldman Sachs: 25.7% chance of winning, Spain far ahead
Goldman Sachs has developed a dedicated statistical model for World Cup predictions. This model records historical data from nearly 20k international A-level matches since 1978, and combines five variables: scoring talent, team momentum, psychological factors, geographical factors, and the advantage of football powerhouses, running 50k Monte Carlo simulations. The results are very clear: Spain leads with a 25.7% chance of winning, France is second at 18.9%, with Argentina, Brazil, and the Netherlands ranking third to fifth. Goldman Sachs' model almost seems to have Spain's championship written on its face.
Opta supercomputer: Spain leads with 16.4%, schedule advantage is obvious
As an authoritative organization in sports data, Opta's supercomputer has run over ten thousand pre-match simulations. The results show Spain leading with a 16.4% probability, followed by France at 12.8%, England at 11.3%, Argentina at 10.4%, and Portugal at 6.9%. Opta's report specifically points out that Spain's schedule is the most favorable among all strong teams—almost certain to advance as the top of their group, with a knockout stage path that is better than other traditional powerhouses. Schedule advantage often outweighs paper strength in knockout tournaments.
Six domestic AI large models: five votes for Spain, one for France
This World Cup is "the first World Cup after humanity entered the AI era," and domestic AI models naturally won't be absent. Media tested six mainstream domestic models—Doubao, Yuanbao, Qianwen, Wenxin Yiyan, Zhipu Qingyan, DeepSeek—and the results are quite interesting: five models—Doubao, Yuanbao, Qianwen, Wenxin Yiyan, Zhipu Qingyan—all bet on Spain, with highly similar reasoning, mainly based on assessments from major authoritative data sources. DeepSeek—the only model voting for France—believes France's luxurious forward line and Mbappé's individual ability are the biggest winning factors.
Kimi: 300-agent cluster bets on Germany "upset"
Among all AI models, Kimi's approach is the most high-profile. It deploys 300 sub-agents working together to analyze all 104 matches, making pre-match predictions and post-match reviews, and even offers a trillion tokens as a prize pool for users to participate in "betting" and share the rewards.
In terms of prediction strategy, Kimi does not follow the crowd entirely. It admits that Spain and France are the biggest favorites, but offers a different view—that Germany's chances might be underestimated by the market, with a possibility of an upset victory. Kimi also openly states, "Our predictions are very likely to be wrong," and places the analysis process, predictions, and post-match reviews within a transparent framework for public verification. As of June 11, the top support teams among Kimi users are: Argentina (43.33%), France, Spain, Brazil, and Portugal.
The emotional appeal of fans versus AI's rationality creates an interesting contrast at this moment.
International AI giants: Gemini and ChatGPT both favor Spain
Looking abroad, Google's Gemini and OpenAI's ChatGPT have also joined the prediction race. Both international AI giants predict Spain will win, aligning with most domestic models. Moreover, the AI predictions for the Golden Boot are surprisingly consistent—pointing entirely to Mbappé. DeepSeek cites Opta data indicating Mbappé is expected to score 6.85 goals, just four goals shy of Klose's all-time World Cup record; Wenxin Yiyan's predicted index for the Golden Boot is even higher at 7.00, leading other contenders by a significant margin.
Championship odds: the market also favors Spain
Betting market odds show the same trend. The latest odds for winning the championship list Spain at 5.5, France at 6.0, England at 7.5, Brazil and Portugal both at 9.0, and defending champion Argentina at 10.0, ranking sixth. The consensus from markets, data, and AI all points to the same answer.
Are AI predictions accurate?
You might be wondering: how reliable are AI predictions? The answer is—highly valuable as a reference, but not to be trusted blindly.
Third-party evaluation platform KellyBench's data shows that top global AI models, including ChatGPT, performed quite poorly in Premier League simulation predictions—many models suffered losses or even "bankruptcy" over the course of the season. Football matches are full of uncertainties like on-the-spot performance, player mentality, sudden injuries, which are difficult for AI to quantify. Industry experts estimate that even the best AI models probably have a success rate of only 60% to 80% in predicting this World Cup. As Hu Yanping, a distinguished professor at Shanghai University of Finance and Economics, said: "Football is round, and the success rate and accuracy of predictions still need to be observed."
