HELM Classic - XSUM: leaderboard
The "XSUM" 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: ROUGE-2 (%). Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 42 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | OPT-175B | 15.49 | #1538 |
| 2 | Palmyra X (43B) | 14.91 | #1317 |
| 3 | text-davinci-002 | 14.37 | #1225 |
| 4 | GLM-130B | 13.24 | #1568 |
| 5 | davinci | 12.63 | #1553 |
| 6 | OPT-66B | 12.62 | #1603 |
| 7 | text-davinci-003 | 12.4 | #1146 |
| 8 | gpt-neox-20B | 10.16 | #1600 |
| 9 | gpt-j-6B | 9.55 | #1609 |
| 10 | text-curie-001 | 7.61 | #1632 |
| 11 | text-babbage-001 | 4.58 | #1648 |
| 12 | text-ada-001 | 3.44 | #1655 |
| 13 | T5 (11B) | 1.46 | #1652 |
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-xsum · How It Works · Data refreshed daily, snapshot 2026-10-20.