HELM v2 Lite - MedQA: leaderboard
The "MedQA" subtask of HELM v2 Lite. Stanford HELM Lite v2 — lightweight HELM run comparing frontier models on core scenarios with exact-match-style metrics.
Metric: Exact Match (%). Source: nlp.stanford.edu. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 (0613) | 88.46 | #491 |
| 2 | GPT-4 Preview (1106) | 86.54 | #418 |
| 3 | Claude v1.3 | 75 | #1001 |
| 4 | Yi 34B (Base) | 73.08 | #972 |
| 5 | Claude 2 | 71.15 | #761 |
| 6 | Claude 2.1 | 69.23 | #839 |
| 7 | Llama 2 70B | 65.38 | #1201 |
| 8 | Yi 6B (Chat) | 61.54 | #1408 |
| 9 | GPT-3.5 Turbo (0613) | 59.62 | #1118 |
| 10 | Mistral-7B-v0.1 | 59.62 | #1326 |
| 11 | text-davinci-003 | 55.77 | #1146 |
| 12 | text-davinci-002 | 55.77 | #1225 |
| 13 | LLaMA-65B | 48.08 | #1303 |
| 14 | Llama 2 7B | 40.38 | #1520 |
| 15 | falcon-40B | 38.46 | #1442 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-v2-lite-medqa · How It Works · Data refreshed daily, snapshot 2026-10-20.