OpenAI HumanEval: leaderboard
OpenAI HumanEval evaluates Python function synthesis from docstrings with executable unit tests.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O4 Mini (2025-04-16) | 98.78 | #179 |
| 2 | Claude 3.7 Sonnet (20250219) | 92.68 | #201 |
| 3 | GPT-4o (2024-08-06) | 92.07 | #336 |
| 4 | Gemini 2.0 Flash (001) | 87.8 | #360 |
| 5 | GPT-4o Mini (2024-07-18) | 87.2 | #558 |
| 6 | DeepSeek V3 Chat | 85.37 | #447 |
| 7 | Qwen 2.5 Coder 32B Instruct | 79.27 | #528 |
| 8 | Llama 3.2 90B Vision Instruct | 76.83 | #622 |
| 9 | Llama 3.3 70B Instruct | 72.56 | #534 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openai-humaneval · How It Works · Data refreshed daily, snapshot 2026-10-20.