CADBench: Gemini 3.7 Flash Ties for First, But Top Pass Rate Only 24.6%

Seldon released CADBench, evaluating 10 models on 105 real Fusion 360 CAD tasks. Google DeepMind's Gemini 3.7 Flash tied with top models for first place at only 24.6% pass rate, suggesting the bottleneck for CAD agents now lies in integration rather than model capability.

2026-08-22 ~ 2026-08-23 · 3 related posts

• Gemini 3.7 Flash Ties for #1 in CAD Computer-Use Benchmark — DynamicWebPaige · 2026-08-22
• CADBench: 10 models tested on 105 Fusion 360 tasks, best pass rate only 24.6% — DynamicWebPaige · 2026-08-22

1 near-duplicate retellings: jakedahn