FutureEval: leaderboard
Bots forecast open Metaculus questions of binary, numeric and multiple-choice type, returning a probability or a full distribution from a fixed prompt with a news search tool. Forecasts are scored once the questions resolve, and the board reports a head-to-head peer skill score built from log-score differences, anchored at zero on a reference model. Higher is better. Source: metaculus.com/futureeval/methodology/.
Metric: Unified Forecasting Score. Source: www.metaculus.com. Saturation forecast: Around July 2027. 97 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 (High) | 16.1 | #16 (Claude Fable 5) |
| 2 | Claude Opus 5 (High) | 15.78 | #12 (Claude Opus 5) |
| 3 | Claude Opus 4.8 (High) | 14.12 | #37 (Claude Opus 4.8) |
| 4 | Gemini 3.6 Flash | 13.22 | #51 |
| 5 | Kimi K3 | 12.93 | #30 |
| 6 | GPT-5.6 Sol (High) | 12.93 | #19 (GPT-5.6 Sol) |
| 7 | Gemini 3.5 Flash | 12.17 | #60 |
| 8 | Gemini 3.1 Pro (Preview) (High) | 11.92 | #58 (Gemini 3.1 Pro (Preview)) |
| 9 | GPT-5.1 (High) | 11.76 | #135 (GPT-5.1) |
| 10 | GPT-5.5 Instant | 11.61 | #129 |
| 11 | O3 | 11.56 | #124 |
| 12 | GPT-5.5 (High) | 11.41 | #31 (GPT-5.5) |
| 13 | GPT-5 | 11.25 | #98 |
| 14 | Claude Opus 4.7 (High) | 11.25 | #49 (Claude Opus 4.7) |
| 15 | GPT-5 (High) | 11.23 | #98 (GPT-5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=futureeval · How It Works · Data refreshed daily, snapshot 2026-10-20.