A leaderboard tracking AI model performance on World Cup 2026 predictions shows Grok 4.3 currently leading with 22 points, highlighting how large language models are now being benchmarked on real-world forecasting tasks.

What happened: An evaluation site has ranked multiple LLMs on their ability to predict FIFA World Cup 2026 outcomes, including match results and tournament picks. Grok 4.3 currently ranks at the top of the leaderboard with 22 points in the active view. The benchmark awards 5 points for correct predictions and 0 for wrong or unresolved ones.

Why it matters: This demonstrates that AI models are being tested against concrete, measurable events rather than abstract benchmarks alone. A model's ability to forecast real sporting tournaments—which have definite outcomes—offers a practical way for business decision-makers to evaluate which AI systems produce reliable predictions in domains that matter to them.

What to watch: The leaderboard tracks 8 model setups across 15 questions, with predictions spanning group winners, semifinalists, top scorer teams, and tournament champion. The results are updated as matches are played, so the rankings will shift as the 2026 World Cup progresses and actual outcomes become known.