HELM v2 Lite - HumanEval (Code) — leaderboard

Metric: Pass@1 (%). Source: nlp.stanford.edu. 18 models tracked.

Top models

#ModelScore
1GPT-4 (0613)73
2text-davinci-00355
3text-davinci-00236
4Yi 34B (Base)30
5PaLM-2 (Bison)29
6LLaMA-65B26
7Llama 2 70B23
8GPT-3.5 Turbo (0613)22
9Mistral-7B-v0.119
10Yi 6B (Chat)16
11Llama 2 13B11
12Llama 2 7B9
13Palmyra X V3 (72B)2
14falcon-7B0
15GPT-4 Preview (1106)0

Interactive version: theaggregate.ai/benchmark?slug=helm-v2-lite-humaneval-code · How the rankings work · Data refreshed daily, snapshot 2026-07-22.