CadQueryEval — Code Execution
Code · 2026-01-10
Percentage of CadQueryEval tasks for which the generated CadQuery Python code executes successfully within the Docker sandbox's 60-second limit.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 100.0 |
| GPT-5.6 Sol Pro | 100.0 |
| GPT-5.5 | 96.0 |
| GPT-5.6 Sol | 96.0 |
| GPT-5.6 Luna | 92.0 |
| GPT-5.6 Luna Pro | 92.0 |
| GPT-5.6 Terra Pro | 92.0 |
| Opus 4.8 | 88.0 |
| DeepSeek-V4-Pro | 88.0 |
| GPT-5.6 Terra | 88.0 |
| Grok 4.5 | 88.0 |
| Kimi K3 | 88.0 |
| Qwen3.7-Max | 88.0 |
| Claude Fable 5 | 84.0 |
| Opus 4.5 | 84.0 |
| Opus 4.6 | 84.0 |
| Sonnet 5 | 84.0 |
| Gemini 3.5 Flash | 80.0 |
| GPT-5 Mini | 80.0 |
| Grok 4.3 | 80.0 |
| Kimi K2.6 | 80.0 |
| MiniMax M3 | 80.0 |
| Muse Spark 1.1 | 80.0 |
| Opus 4.7 | 76.0 |
| Sonnet 4.6 | 76.0 |
| Gemini 3 Pro Preview | 76.0 |
| GLM-5.1 | 76.0 |
| GLM-5.2 | 76.0 |
| Kimi K2.5 | 76.0 |
| Qwen3.7-Plus | 76.0 |
| Gemini 3 Flash Preview | 72.0 |
| GPT-5 | 72.0 |
| GPT-5.4 | 72.0 |
| Opus 4 | 68.0 |
| Sonnet 4 | 68.0 |
| GPT-5.1 | 68.0 |
| GPT-5.2 | 68.0 |
| O4 Mini | 68.0 |
| Opus 4.1 | 64.0 |
| Hy3 preview | 64.0 |