- Alibaba’s football forecasting tool correctly calls opening two matches
- Model adds red cards, weather and altitude into match predictions
Alibaba Group’s Qwen AI football prediction assistant has become an early talking point at the 2026 FIFA World Cup after correctly forecasting the outcomes of the tournament’s first two matches and even anticipating key in-game events.
The tool predicted Mexico’s 2-0 win over South Africa and South Korea’s 2-1 victory over the Czech Republic on opening day, both of which materialized, prompting online comparisons to a “digital Paul the Octopus.”
“Digital Paul the Octopus” is a nickname referring to AI systems being compared to the famous octopus Paul from the 2010 World Cup, which became a viral sensation for correctly predicting match outcomes.
Correcting predicting red cards
In the opening match in Mexico City on June 12 (Beijing time), Qwen forecast a 2-0 win for Mexico and flagged the refereeing style of Brazilian official Wilton Sampaio as “strict,” warning that South Africa could be at risk of disciplinary issues if defensive intensity crossed a threshold.
The prediction proved unusually specific. South Africa were reduced to 10 men in the 49th minute, and the match produced three red cards in total, a record for a World Cup opener. Mexico striker Raúl Jiménez, highlighted in Qwen’s analysis, also scored.
The second match between South Korea and the Czech Republic drew further attention after multiple competing AI systems, including OpenAI’s GPT and Google’s Gemini, reportedly leaned toward an unbeaten Czech outcome. Qwen, however, predicted a 2-1 win for the South Koreans.
Firm on the South Korea game scoreline
During the match, with the score level at 1-1, users asked whether the model would revise its forecast. Qwen responded that it would not change its view and maintained that South Korea would score again.
A decisive goal followed in the 80th minute to secure the 2-1 result.
Qwen AI prediction product lead Cheng Fei said earlier the system is trained on large-scale datasets that combine historical team performance and player injury records with environmental variables such as altitude, terrain, temperature and humidity.
In the opening match, the model specifically cited Mexico City’s altitude of more than 2,200 meters as a factor favoring the host team.
Morocco beating Brazil?
The tool has now made a more controversial call for its next fixture, predicting that Morocco will upset Brazil.
As of 13:00 on June 13, 71% of online users expected a Brazil win, while only 20% backed Morocco, setting up a high-profile test of human versus machine forecasting.
