BenchCAD - geometric overlap
Multimodal
Reconstruct CAD objects from multi-view renders and measure geometric overlap (IoU). Tool availability is recorded per score; exact task count is not disclosed by the release tables.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 96.2 |
| GPT-6 Astra | 95.9 |
| Kimi K3.1 Pro | 92.7 |
| Claude Fable 5.1 | 92.6 |
| Claude Opus 5 | 89.9 |
| GPT-5.6 Sol | 83.4 |
| GPT-5.6 Terra | 78.2 |
| GPT-5.6 Luna | 73.9 |