HELM Classic - NaturalQuestions Open Book: leaderboard
The "NaturalQuestions Open Book" subtask of HELM Classic. A scenario from Stanford HELM Classic, the holistic evaluation suite scoring language models across accuracy, robustness and calibration on a broad set of tasks.
Metric: F1 (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 66 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | text-davinci-003 | 77.02 | #1146 |
| 2 | text-davinci-002 | 71.32 | #1225 |
| 3 | mpt-30B Instruct | 69.72 | #1470 |
| 4 | Mistral-7B-v0.1 | 68.66 | #1326 |
| 5 | vicuna-13B-v1.3 | 68.65 | #1447 |
| 6 | falcon-40B | 67.53 | #1442 |
| 7 | GPT-3.5 Turbo (0613) | 67.48 | #1118 |
| 8 | Llama 2 70B | 67.42 | #1201 |
| 9 | mpt-30B | 67.29 | #1476 |
| 10 | LLaMA-65B | 67.21 | #1303 |
| 11 | falcon-40B Instruct | 66.59 | #1458 |
| 12 | LLaMA-30B | 66.56 | #1423 |
| 13 | RedPajama-INCITE-7B-Instruct | 65.92 | #1566 |
| 14 | GLM-130B | 64.24 | #1568 |
| 15 | Llama 2 13B | 63.73 | #1436 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-classic-naturalquestions-open-book · How It Works · Data refreshed daily, snapshot 2026-10-20.