MMLU-by-task - Public Relations — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 1257 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | LLaMA-65B | 74.55 |
| 2 | Llama 2 70B Chat | 74.55 |
| 3 | falcon-180B | 73.64 |
| 4 | LLaMA-30B | 70 |
| 5 | Llama 2 70B Chat GPTQ | 70 |
| 6 | Llama 2 70B Base | 69.09 |
| 7 | Llama 2 70B Chat (HF) | 69.09 |
| 8 | StableBeluga2 | 68.18 |
| 9 | vicuna-13B-v1.5 | 68.18 |
| 10 | Mistral-7B-v0.1 | 67.27 |
| 11 | MythoMax-L2-13B | 67.27 |
| 12 | WizardLM-13B-V1.2 | 67.27 |
| 13 | Llama 2 13B Chat Base | 66.36 |
| 14 | falcon-40B Instruct | 66.36 |
| 15 | OpenHermes-13B | 65.45 |
Interactive version: theaggregate.ai/benchmark?slug=mmlu-by-task-public-relations · How the rankings work · Data refreshed daily, snapshot 2026-07-22.