BenchCAD 2.1 (Anthropic runs): leaderboard
Models write programmatic CAD for the 100 held-out BenchCAD 2.1 Core cases across six tasks (parts and assemblies from engineering drawings or rendered views, board wiring from renders); the score (0 to 1) is the mean of 5 passes when Anthropic ran each Claude model in BenchCAD's standard 2.1 harness (60 rounds, 500,000 output tokens per case). Unlike the BenchCAD figures in OpenAI's and Kimi's launch posts, which the BenchCAD site files under its 1.0 Vision2Code voxel-IoU task, this is the 2.1 Core set, with Anthropic's runs of four Claude models at five efforts.
Metric: Core score (0 to 1) on the 100 held-out BenchCAD 2.1 Core cases (six programmatic-CAD tasks), standard 2.1 harness (60 rounds and 500,000 output tokens per case, reasoning carried across rounds), mean of 5 passes; Anthropic's runs. Source: benchcad.com. Saturation forecast: Around December 2027. 20 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 (Max) | 0.49 | #5 (Claude Opus 5.5) |
| 2 | Claude Sonnet 5.5 (Max) | 0.47 | #19 (Claude Sonnet 5.5) |
| 3 | Claude Opus 5.5 (xHigh) | 0.42 | #5 (Claude Opus 5.5) |
| 4 | Claude Sonnet 5.5 (xHigh) | 0.37 | #19 (Claude Sonnet 5.5) |
| 5 | Claude Fable 5.1 (Max) | 0.36 | #15 (Claude Fable 5.1) |
| 6 | Claude Fable 5.1 (xHigh) | 0.34 | #15 (Claude Fable 5.1) |
| 7 | Claude Opus 5.5 (High) | 0.24 | #5 (Claude Opus 5.5) |
| 8 | Claude Fable 5.1 (High) | 0.2 | #15 (Claude Fable 5.1) |
| 9 | Claude Haiku 5.5 (Max) | 0.19 | #65 (Claude Haiku 5.5) |
| 10 | Claude Sonnet 5.5 (High) | 0.16 | #19 (Claude Sonnet 5.5) |
| 11 | Claude Opus 5.5 (Medium) | 0.15 | #5 (Claude Opus 5.5) |
| 12 | Claude Haiku 5.5 (xHigh) | 0.15 | #65 (Claude Haiku 5.5) |
| 13 | Claude Fable 5.1 (Medium) | 0.12 | #15 (Claude Fable 5.1) |
| 14 | Claude Haiku 5.5 (High) | 0.09 | #65 (Claude Haiku 5.5) |
| 15 | Claude Fable 5.1 (Low) | 0.08 | #15 (Claude Fable 5.1) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #6 GPT-6 Astra.
Interactive version: theaggregate.ai/benchmark?slug=benchcad-2-1-anthropic-runs · How It Works · Data refreshed daily, snapshot 2026-11-13.