Benchmark
Parametric CAD Bench v3
Scores are percentages.
Benchmark sourceScores from Harbor Hub Leaderboards.
- Models measured
- 11
- Scores recorded
- 11
- Sources
- 1
- IQ weight
- 0.92
Score distribution
- 37.9%
- Mean
- 34.8%
- Median
- 15%
- Std dev
- 20.9%
- 10th percentile
- 56.9%
- 90th percentile
- 17.2%
- Min
This benchmark has a weight of 0.92 for chat models, measured over 11 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | |
|---|---|---|
| 1 | Claude Opus 5.5v3· Harbor Hub Leaderboards | 61% |
| 2 | GPT-6 Astrav3· Harbor Hub Leaderboards | 56.9% |
| 3 | Claude Fable 5.1v3· Harbor Hub Leaderboards | 56.8% |
| 4 | Claude Opus 5v3· Harbor Hub Leaderboards | 50.9% |
| 5 | Gemini 3.8 Flashv3· Harbor Hub Leaderboards | 36.2% |
| 6 | Grok 4.7v3· Harbor Hub Leaderboards | 34.8% |
| 7 | GPT-5.6 Solv3· Harbor Hub Leaderboards | 30% |
| 8 | Grok 4.6v3· Harbor Hub Leaderboards | 27.7% |
| 9 | Kimi K3v3· Harbor Hub Leaderboards | 24.9% |
| 10 | Muse Spark 1.3v3· Harbor Hub Leaderboards | 20.9% |
| 11 | GPT-5.6 Terrav3· Harbor Hub Leaderboards | 17.2% |
Tokens per task, input and output together, as measured by the source. Comparable within this benchmark only.