Vector Eval - HumanEval: leaderboard
Vector Institute Inspect Evals run of HumanEval, measuring Python code-generation correctness on function synthesis problems.
Metric: Mean Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 Mini | 98.17 | #272 |
| 2 | O1 | 96.95 | #183 |
| 3 | DeepSeek R1 | 95.68 | #251 |
| 4 | Claude 3.5 Sonnet | 94.51 | #350 |
| 5 | Qwen3 Coder Next | 91.46 | #326 |
| 6 | GPT-4o | 90.85 | #343 |
| 7 | Gemini 1.5 Pro | 87.2 | #445 |
| 8 | Mistral Large 2 (Jul) | 86.59 | #484 |
| 9 | GPT-4o Mini | 85.98 | #599 |
| 10 | Llama 3.1 70B Instruct | 78.66 | #561 |
| 11 | Gemini 1.5 Flash | 74.39 | #618 |
| 12 | Command-R+ | 62.2 | #891 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vector-eval-humaneval · How It Works · Data refreshed daily, snapshot 2026-10-20.