HELM Classic - Synthetic Reasoning Abstract: leaderboard
The "Synthetic Reasoning Abstract" 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. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-3.5 Turbo (0613) | 50.94 | #1118 |
| 2 | Palmyra X (43B) | 50.36 | #1317 |
| 3 | text-davinci-003 | 50.25 | #1146 |
| 4 | Llama 2 70B | 48.87 | #1201 |
| 5 | text-davinci-002 | 48.78 | #1225 |
| 6 | Mistral-7B-v0.1 | 45.11 | #1326 |
| 7 | LLaMA-65B | 43.82 | #1303 |
| 8 | LLaMA-30B | 41.23 | #1423 |
| 9 | mpt-30B Instruct | 34.95 | #1470 |
| 10 | mpt-30B | 33.53 | #1476 |
| 11 | Llama 2 13B | 32.36 | #1436 |
| 12 | GLM-130B | 25.16 | #1568 |
| 13 | falcon-40B | 23.75 | #1442 |
| 14 | davinci | 23.58 | #1553 |
| 15 | vicuna-13B-v1.3 | 23.37 | #1447 |
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-synthetic-reasoning-abstract · How It Works · Data refreshed daily, snapshot 2026-10-20.