Seven AI Models Forecast 2026 World Cup; Four Pick Spain, Three Choose Argentina

Seven artificial intelligence models tasked with forecasting the 2026 FIFA World Cup produced a split verdict, with four selecting Spain and three choosing Argentina as the likely champion, according to a published experiment that ran each system through the full 48-team, 12-group draw and knockout bracket. The models, configured as "Hermes agents" with access to free statistics sites and custom analytical skills, were given identical inputs and freedom to choose their own forecasting methods.

Anthropic's Opus 4.8 Max, described as treating the tournament "like a physics problem," used Dixon-Coles Poisson modeling combined with Monte-Carlo bracket simulation and emerged with Spain as champion at a 20% probability, defeating France in the final, with Portugal and England losing in the semifinals. The model was the only one in the field to incorporate off-field variables such as severe heat across approximately five matches, the 2,200-meter altitude at Estadio Azteca, and continent-sized travel, treating those conditions as a structural advantage for deeper European squads. Opus cut Brazil's odds to 8%, roughly half what Argentina-leaning models assigned, citing Rodrygo's knee injury, Estêvão's fitness concerns, and a 34-year-old Neymar recalled for the tournament. Its headline matchup was a quarterfinal billed as "the real final, a round early," with Spain eliminating a 39-year-old Lionel Messi and Argentina, and it selected Kylian Mbappé for the Golden Boot.