OmniMech - CAD Program Pass Rate (Drawing + 8 Renders): leaderboard
OmniMech (2026) CAD program synthesis with the 2D engineering drawing plus eight multi-view renderings of the part as input: the percentage of 5,000 sampled drawings for which the model returns a complete program that builds a valid, non-empty 3D model. Higher is better.
Metric: Pass rate (%; input: the drawing plus 8 multi-view renderings of the part; 5,000 sampled industrial 2D orthographic drawings with dimension and tolerance annotations, zero-shot; a trial passes only when the model returns a complete response whose extracted CAD program builds a valid, non-empty 3D model). Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 | 94.2 | #56 |
| 2 | GPT-5.5 | 91.5 | #49 |
| 3 | Gemini 3.1 Pro (Preview) | 82.3 | #94 |
| 4 | Qwen 3.7 Plus | 71.2 | #207 |
| 5 | Gemini 3.5 Flash | 68 | #101 |
| 6 | GPT-5.4 Mini | 53 | #389 |
| 7 | Llama 4 Maverick | 45.9 | #1099 |
| 8 | Qwen 3.5 9B | 1 | #803 |
| 9 | Ministral 3 14B | 0.6 | #1845 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=omnimech-cad-program-pass-rate-drawing-plus-8-renders · How It Works · Data refreshed daily, snapshot 2026-11-13.