ForecastBench Tournament Market Brier Index
Forecasting · 2024-09-30
Difficulty-adjusted Brier Index on ForecastBench market questions.
Top models (higher is better)
| Model | Score |
|---|---|
| Grok 4.20 | 72.6 |
| Claude Opus 5 | 72.3 |
| GPT-5.6 Sol | 72.2 |
| Opus 4.8 | 72.0 |
| GPT-5.6 Sol Pro | 71.7 |
| Grok 4.5 | 71.7 |
| GPT-5.5 | 71.6 |
| Opus 4.5 | 70.9 |
| Grok 4.6 | 70.9 |
| Gemini 3.6 Flash | 70.8 |
| Gemini 3.7 Flash | 70.8 |
| Grok 4 | 70.8 |
| Gemini 3 Pro Preview | 70.7 |
| Grok Beta | 70.7 |
| GPT-4.5 | 70.6 |
| Grok 4 Fast | 70.6 |
| Gemini 3.1 Pro Preview | 70.5 |
| Sonnet 4 | 70.4 |
| Claude 3.7 Sonnet | 70.4 |
| GPT-5 | 70.3 |
| Grok 4.1 Fast | 70.2 |
| Opus 4.1 | 69.9 |
| Gemini 3 Flash Preview | 69.9 |
| Claude 3.5 Sonnet (Oct 2024) | 69.8 |
| o3-mini | 69.7 |
| o3 | 69.6 |
| GPT-4 Turbo | 69.5 |
| Kimi K3 | 69.5 |
| Gemini 2.5 Pro | 69.3 |
| MiniMax M3 | 69.3 |
| Qwen3-235B-A22B | 69.3 |
| GPT-5.2 | 69.2 |
| Sonnet 4.5 | 69.1 |
| Sonnet 4.6 | 69.1 |
| DeepSeek-V3 | 69.0 |
| Haiku 4.5 | 68.9 |
| QwQ-32B-Preview | 68.9 |
| DeepSeek-R1 | 68.7 |
| Gemini 3.5 Flash | 68.7 |
| GLM 4.5 Air | 68.6 |