ForecastBench Baseline Dataset Brier Index
Forecasting · 2024-09-30
Difficulty-adjusted Brier Index on baseline dataset questions.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 2.5 Pro Preview 03-25 | 61.4 |
| GPT-5 | 61.4 |
| GPT-4.5 | 61.3 |
| Gemini 2.5 Flash Preview 04-17 | 60.9 |
| Gemini 3 Pro Preview | 60.8 |
| O4 Mini | 60.8 |
| Opus 4 | 60.6 |
| o3 | 60.6 |
| Opus 4.7 | 60.5 |
| Gemini 3.1 Pro Preview | 60.5 |
| GPT-4.1 | 60.4 |
| Opus 4.1 | 60.3 |
| GPT-5.5 | 60.3 |
| GPT-5 Mini | 60.3 |
| Grok 4 | 60.3 |
| Grok 4.1 Fast | 60.3 |
| Grok 4.3 | 60.3 |
| Grok 4 Fast | 60.3 |
| Claude 3.7 Sonnet | 60.3 |
| Gemini 3.5 Flash | 60.2 |
| Opus 4.5 | 60.1 |
| Sonnet 4.5 | 60.1 |
| Sonnet 4.6 | 60.1 |
| Gemini 2.5 Pro | 60.1 |
| Opus 4.6 | 60.0 |
| Gemini 3 Flash Preview | 60.0 |
| GLM-5 | 60.0 |
| GPT-5.2 | 59.9 |
| GPT-5.4 | 59.9 |
| Grok 4.20 | 59.9 |
| Qwen3-235B-A22B | 59.8 |
| Gemini 3.1 Flash-Lite Preview | 59.6 |
| Haiku 4.5 | 59.1 |
| Sonnet 4 | 59.1 |
| Gemini 3.1 Flash-Lite | 59.1 |
| DeepSeek-R1 | 59.0 |
| o3-mini | 59.0 |
| DeepSeek-V3 | 58.9 |
| Claude 3.5 Sonnet (Oct 2024) | 58.8 |
| Kimi K2 Instruct 0905 | 58.8 |