The stock of Knowledge Atlas (02513.HK) surged dramatically on June 18th, opening higher and continuing its ascent to close at HK$2,094, marking a significant 26.14% gain from the previous session and reaching a new historical peak. This rally has drawn widespread market attention to the broader AI industry chain.

Concurrently, the ongoing 2026 FIFA World Cup has become a key platform for showcasing the capabilities of large language models. In the "World Cup Prediction: AI vs Human" contest, a joint initiative by Lenovo Group and Migu Video, Knowledge Atlas currently holds a 50% prediction accuracy rate after 24 matches, placing it within the third tier of competitors.

In the overall standings, Baidu's ERNIE model leads with a 58.3% win rate. The second tier, with a 54.2% accuracy, is shared by Lenovo's Tianxi AI, China Mobile's Jiutian, and Tencent's Hunyuan. Following closely behind are models like Knowledge Atlas, DeepSeek, and Tongyi Qianwen, all with a 50% success rate.

This interactive World Cup event, the first of its kind to pit AI collectives against the general public in prediction contests, has gathered 12 leading domestic AI models to provide continuous forecasts for all 104 tournament matches. While the AI models have collectively stumbled in predicting some major upsets, the competition offers a valuable real-world testing ground for these large language models.

Industry observers note that as AI capabilities increasingly transition from research labs to practical applications like sports, content creation, and event interaction, the commercial potential of these models is attracting greater scrutiny from capital markets. The World Cup, as a premier global sporting event, is becoming a crucial window for evaluating the real-world implementation and effectiveness of AI technologies.