HELM Classic - CivilComments: leaderboard
The "CivilComments" 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: Exact Match (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 67 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-3.5 Turbo (0613) | 69.64 | #1118 |
| 2 | text-davinci-003 | 68.44 | #1146 |
| 3 | text-davinci-002 | 66.84 | #1225 |
| 4 | RedPajama-INCITE-7B-Instruct | 66.45 | #1566 |
| 5 | LLaMA-65B | 65.46 | #1303 |
| 6 | Llama 2 70B | 65.19 | #1201 |
| 7 | vicuna-13B-v1.3 | 64.46 | #1447 |
| 8 | Mistral-7B-v0.1 | 62.44 | #1326 |
| 9 | vicuna-7B-v1.3 | 62.04 | #1507 |
| 10 | falcon-40B Instruct | 60.27 | #1458 |
| 11 | LLaMA-13B | 59.98 | #1516 |
| 12 | mpt-30B | 59.9 | #1476 |
| 13 | Llama 2 13B | 58.77 | #1436 |
| 14 | mpt-30B Instruct | 57.31 | #1470 |
| 15 | alpaca-7B | 56.58 | #1611 |
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-civilcomments · How It Works · Data refreshed daily, snapshot 2026-10-20.