Forecastbench — Epoch AI
Forecastbench — Epoch AI is run by Epoch AI. It has ranked 81 models, of which the catalogue holds 65, scoring from 50.4 to 62.5 on Overall score.
measured by Epoch AI · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 81 | o3 | Overall score | 62.5 |
| 2ndof 81 | Claude Opus 4.1 | Overall score | 62 |
| 3rdof 81 | Claude Sonnet 4.5 | Overall score | 61.9 |
| 4thof 81 | Claude 3.7 Sonnet | Overall score | 61.8 |
| 5thof 81 | o4-mini | Overall score | 61.8 |
| 7thof 81 | GPT-4.1 fine-tuned | Overall score | 61.5 |
| 8thof 81 | Claude Haiku 4.5 | Overall score | 61.4 |
| 9thof 81 | Grok 4.20 | Overall score | 61.4 |
| 10thof 81 | GPT-5 | Overall score | 61.4 |
| 11thof 81 | Gemini 2.5 Pro | Overall score | 61.3 |
| 12thof 81 | gemini-3-pro-preview | Overall score | 61.2 |
| 13thof 81 | Claude Sonnet 4.6 | Overall score | 61.2 |
| 14thof 81 | Claude Opus 4 | Overall score | 61.1 |
| 15thof 81 | GPT-5.5 | Overall score | 61.1 |
| 16thof 81 | Grok 4.1 Fast Reasoning | Overall score | 61 |
| 17thof 81 | GPT-5 mini | Overall score | 61 |
| 18thof 81 | GLM-5 | Overall score | 61 |
| 20thof 81 | Claude Opus 4.5 | Overall score | 60.7 |
| 21stof 81 | Claude 3.5 Sonnet | Overall score | 60.7 |
| 23rdof 81 | Gemini 2.5 Flash | Overall score | 60.6 |
| 24thof 81 | Grok 4 Fast | Overall score | 60.5 |
| 26thof 81 | Grok 4.3 | Overall score | 60.4 |
| 27thof 81 | MiniMax M3 | Overall score | 60.4 |
| 28thof 81 | Kimi K2 Instruct | Overall score | 60.2 |
| 29thof 81 | Claude Sonnet 4 | Overall score | 60.2 |
| 30thof 81 | GPT-5.2 | Overall score | 60.1 |
| 31stof 81 | Claude Opus 4.6 | Overall score | 60 |
| 32ndof 81 | DeepSeek-R1 | Overall score | 60 |
| 33rdof 81 | Meta-Llama-3.1-405B-Instruct | Overall score | 59.9 |
| 34thof 81 | Claude Sonnet 5 | Overall score | 59.9 |
| 36thof 81 | Qwen3 235B A22B | Overall score | 59.7 |
| 37thof 81 | o3-mini | Overall score | 59.6 |
| 39thof 81 | Claude Opus 4.7 | Overall score | 59.6 |
| 40thof 81 | GPT-5.4 | Overall score | 59.5 |
| 41stof 81 | GPT-4 Turbo | Overall score | 59.4 |
| 43rdof 81 | Claude Opus 4.8 | Overall score | 59.3 |
| 44thof 81 | GLM-4.5-Air | Overall score | 59.2 |
| 45thof 81 | Gemini 3.5 Flash | Overall score | 59.2 |
| 46thof 81 | GPT-5 nano | Overall score | 59.1 |
| 47thof 81 | DeepSeek-V3 | Overall score | 59.1 |
| 48thof 81 | Gemini 3.1 Pro Preview | Overall score | 59 |
| 50thof 81 | Llama 3.3 70B Instruct | Overall score | 58.6 |
| 51stof 81 | Llama 3 8B Instruct | Overall score | 58.6 |
| 52ndof 81 | Gemini 3 Flash Preview | Overall score | 58.5 |
| 55thof 81 | Claude 3 Opus | Overall score | 58.4 |
| 56thof 81 | QwQ-32B-Preview | Overall score | 58.3 |
| 57thof 81 | GPT-5.1 | Overall score | 58.1 |
| 58thof 81 | DeepSeek-V3.1 | Overall score | 58 |
| 59thof 81 | GPT-4 (0613) | Overall score | 57.8 |
| 60thof 81 | GPT-4o (2024-05-13) | Overall score | 57.7 |
| 63rdof 81 | Qwen2.5 72B Instruct | Overall score | 57.5 |
| 64thof 81 | Llama 4 Maverick 17B | Overall score | 57.5 |
| 65thof 81 | GPT-5.4 nano | Overall score | 57.3 |
| 66thof 81 | Mistral Large 2407 | Overall score | 57.1 |
| 67thof 81 | Llama 3 70B | Overall score | 57.1 |
| 69thof 81 | GPT-5.4 mini | Overall score | 57 |
| 71stof 81 | Mixtral-8x7B-Instruct-v0.1 | Overall score | 56.3 |
| 72ndof 81 | Mixtral-8x22B-Instruct-v0.1 | Overall score | 56.3 |
| 74thof 81 | DeepSeek V4 Pro | Overall score | 55.3 |
| 75thof 81 | Gemini 3.1 Flash-Lite | Overall score | 54.4 |
| 76thof 81 | Claude 2.1 | Overall score | 54.2 |
| 78thof 81 | Claude 3 Haiku | Overall score | 53.2 |
| 79thof 81 | Llama 3 8B | Overall score | 52.9 |
| 80thof 81 | Llama-2-70b-chat-hf | Overall score | 51.4 |
| 81stof 81 | GPT-3.5 Turbo | Overall score | 50.4 |
Read from the board on 2026-09-16