ForecastBench Baseline - Zero-Shot - Dataset: leaderboard
ForecastBench Baseline leaderboard, where the ForecastBench team runs models out of the box, zero-shot prompt. Dataset component: questions generated automatically from time-series data sources; difficulty-adjusted Brier Index on 0-100 (higher is better; 50 equals always forecasting 0.5). Superforecaster and public medians, naive baselines and LLM crowds are excluded.
Metric: Dataset Brier Index (difficulty-adjusted, higher is better). Source: www.forecastbench.org. Saturation forecast: Around 2031. 81 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 61.2 | #126 |
| 2 | GPT-5 | 61 | #105 |
| 3 | GPT-5.5 (2026-04-23) | 60.7 | #25 |
| 4 | Gemini 3.1 Pro (Preview) | 60.6 | #64 |
| 5 | O4 Mini (2025-04-16) | 60.6 | #183 |
| 6 | Gemini 3 Pro (Preview) | 60.5 | #71 |
| 7 | Gemini 3.5 Flash | 60.5 | #66 |
| 8 | Gemini 2.5 Flash (Preview 04-17) | 60.5 | #246 |
| 9 | Kimi K3 (Max) | 60.4 | #35 (Kimi K3) |
| 10 | Grok 4.20 0309 (Reasoning) | 60.4 | #120 (Grok 4.20 0309) |
| 11 | Gemini 3.1 Pro (Preview) (High) | 60.3 | #64 (Gemini 3.1 Pro (Preview)) |
| 12 | O3 (2025-04-16) | 60.2 | #125 |
| 13 | Gemini 3 Flash (Preview) | 60.1 | #90 |
| 14 | Grok 4.1 Fast (Reasoning) | 60.1 | #227 (Grok 4.1 Fast) |
| 15 | Grok 4.20 Beta (0309) (Reasoning) | 60.1 | #198 (Grok 4.20 Beta (0309)) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=forecastbench-baseline-zero-shot-dataset · How It Works · Data refreshed daily, snapshot 2026-11-05.