BenchCAD Vision2Code voxel IoU without tools (BenchCAD Vision2Code (no tools))
We show this table for reference; we do not rank on it.
Generates CadQuery code from multi-view renders and scores geometric similarity by voxel intersection-over-union.
Benchmark score on BenchCAD Vision2Code (no tools) — November 13, 2026
We compile the BenchCAD Vision2Code (no tools) rows from provider self-reports. Claude Sonnet 5.5 leads the table at 0.747, followed by Claude Opus 5.5 (0.730) and Kimi K3.1 Pro (0.728). We do not use these results to rank models overall.
Claude Sonnet 5.5
Anthropic
Reported by AnthropicClaude Opus 5.5
Anthropic
Reported by AnthropicKimi K3.1 Pro
Moonshot AI
Reported by Moonshot AI5 modelsMultimodal & GroundedCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (5 models)
All 5 results on this page were reported by the labs that built the models.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 5.5AnthropicReported by Anthropic | 0.747 | Reported by Anthropic | Not reported | Closed |
| 2 | Claude Opus 5.5AnthropicReported by Anthropic | 0.730 | Reported by Anthropic | Not reported | Closed |
| 3 | Kimi K3.1 ProMoonshot AIReported by Moonshot AI | 0.728 | Reported by Moonshot AI | Not reported | Open |
| 4 | Claude Fable 5.1AnthropicReported by Anthropic | 0.437 | Reported by Anthropic | Not reported | Closed |
| 5 | Claude Opus 5AnthropicReported by Anthropic | 0.366 | Reported by Anthropic | Not reported | Closed |
- Lab-reported
- Benchmark-run
- Independent
Among the reported BenchCAD Vision2Code (no tools) rows, Claude Sonnet 5.5 is first at 0.747. The third row is 0.019 score units behind. The broader top-10 range is 0.381 score units, so many of the published results sit in a relatively narrow band.
5 models have been evaluated on BenchCAD Vision2Code (no tools). The benchmark falls in the Multimodal & Grounded category. BenchCAD Vision2Code (no tools) is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About BenchCAD Vision2Code (no tools)
Year
2026
Tasks
1,000-file Vision2Code subset
Format
Voxel IoU
Difficulty
Programmatic CAD generation
Section 8.12.2 reports the five-run average on a random 1,000-file subset with max effort and no tools. The card documents two implementation fixes already merged upstream.
Freshness and provenance
Version
BenchCAD Vision2Code (no tools) 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does BenchCAD Vision2Code (no tools) measure?
Generates CadQuery code from multi-view renders and scores geometric similarity by voxel intersection-over-union.
Which model scores highest on BenchCAD Vision2Code (no tools)?
Claude Sonnet 5.5 by Anthropic currently leads with a score of 0.747 on BenchCAD Vision2Code (no tools).
How many models are evaluated on BenchCAD Vision2Code (no tools)?
5 AI models have published results on BenchCAD Vision2Code (no tools) in the BenchLM catalog.
Compare top models on BenchCAD Vision2Code (no tools)
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.