WC2026-Agents: leaderboard
WC2026-Agents (University of Memphis and QuantaInsight, 2026) evaluates assistants with web search as forecasting agents on the 104 matches of the 2026 FIFA World Cup, each forecast about a day before kickoff. The score is the multiclass Brier score of the win, draw and loss probabilities, from 0 to 2; lower is better.
Metric: Multiclass Brier score of the team-A win, draw and team-B win probabilities committed about 24 hours before kickoff for all 104 matches of the 2026 FIFA World Cup (0 to 2), with web search in the assistant consumer interface; lower is better. Source: arxiv.org. Saturation forecast: Around February 2027. 4 models tracked. Updated 30 Sep 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Claude Opus 4.8 | 0.4705 | #58 | |
| 2 | WC2026-Agents Grok Expert Mode (checkpoint unspecified) | 0.4706 | ||
| 3 | GPT-5.5 | High | 0.4729 | #52 |
| 4 | Gemini 3.1 Pro (Preview) | 0.4828 | #93 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=wc2026-agents · How It Works · Data refreshed daily, snapshot 2026-11-16.