WorldCupBench: leaderboard
Live 2026 World Cup forecasting benchmark where frontier models submitted frozen tournament predictions and are scored by match-prediction points and Brier score as results arrive.
Metric: Quiniela Points. Source: github.com. Saturation forecast: Not forecast. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 | 110 | #16 |
| 2 | DeepSeek V4 Pro | 90 | #102 |
| 3 | MiniMax-M3 | 87 | #136 |
| 4 | Gemini 3.5 Flash | 81 | #60 |
| 5 | Grok 4.3 | 77 | #188 |
| 6 | MiMo-V2.5-Pro | 76 | #158 |
| 7 | GLM-5.1 | 74 | #97 |
| 8 | GPT-5.5 | 71 | #31 |
| 9 | Qwen 3.7 Max | 70 | #75 |
| 10 | Kimi K2.6 | 69 | #104 |
| 11 | Nex N2 Pro | 67 | #107 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=worldcupbench · How It Works · Data refreshed daily, snapshot 2026-10-20.