HELM v2 Lite - HumanEval (Code): leaderboard
The "HumanEval (Code)" subtask of HELM v2 Lite. Stanford HELM Lite v2 — lightweight HELM run comparing frontier models on core scenarios with exact-match-style metrics.
Metric: Pass@1 (%). Source: nlp.stanford.edu. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 (0613) | 73 | #491 |
| 2 | text-davinci-003 | 55 | #1146 |
| 3 | text-davinci-002 | 36 | #1225 |
| 4 | Yi 34B (Base) | 30 | #972 |
| 5 | LLaMA-65B | 26 | #1303 |
| 6 | Llama 2 70B | 23 | #1201 |
| 7 | GPT-3.5 Turbo (0613) | 22 | #1118 |
| 8 | Mistral-7B-v0.1 | 19 | #1326 |
| 9 | Yi 6B (Chat) | 16 | #1408 |
| 10 | Llama 2 13B | 11 | #1436 |
| 11 | Llama 2 7B | 9 | #1520 |
| 12 | falcon-7B | 0 | #1598 |
| 13 | GPT-4 Preview (1106) | 0 | #418 |
| 14 | falcon-40B | 0 | #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-humaneval-code · How It Works · Data refreshed daily, snapshot 2026-10-20.