HELM MMLU - Conceptual Physics: leaderboard
The "Conceptual Physics" subtask of HELM MMLU. HELM Classic MMLU group leaderboard from the public HELM release artifacts.
Metric: EM. Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 79 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 1.5 Pro (001) | 94.89 | #446 |
| 2 | Gemini 1.5 Pro (002) | 94.47 | #350 |
| 3 | DeepSeek V3 | 94.04 | #323 |
| 4 | GPT-4o (2024-08-06) | 92.34 | #335 |
| 5 | GPT-4o (2024-05-13) | 91.06 | #326 |
| 6 | Claude 3.5 Sonnet (20241022) | 90.64 | #291 |
| 7 | GPT-4 Turbo | 89.36 | #406 |
| 8 | GPT-4 Turbo (Preview) | 89.36 | #399 |
| 9 | Qwen 2.5 72B Instruct | 88.51 | #458 |
| 10 | Claude 3.5 Sonnet (20240620) | 88.51 | #298 |
| 11 | Claude 3 Opus (20240229) | 88.09 | #413 |
| 12 | Llama 3.1 405B Instruct | 87.66 | #461 |
| 13 | Qwen 2 72B Instruct | 87.23 | #579 |
| 14 | GPT-4 (0613) | 86.81 | #494 |
| 15 | Mistral Large 2 (Jul) | 86.38 | #485 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-mmlu-conceptual-physics · How It Works · Data refreshed daily, snapshot 2026-10-23.