Text2CAD-Bench: 600 Human-Curated Examples for Text-to-CAD Generation

Text2CAD-Bench is a benchmark for evaluating text-to-CAD generation, comprising 600 human-curated examples organized into four levels. It was created by AICAD and released in February 2026. The benchmark is designed to assess performance across geometric complexity and application diversity.

Use Cases

• Benchmarking text-to-CAD model performance based on the described four levels of geometric complexity.
• Evaluating AI-generated CAD model diversity based on the described application categories.
• Training or fine-tuning generative models for CAD using the human-curated text-CAD example pairs.

Strengths

• Contains 600 human-curated examples for evaluation.
• Structured into four benchmark levels to assess different aspects of generation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-02-06 03:58:06; freshness should be verified.

Provenance

Source

AICAD via Hugging Face.

Collection Method

Human-curated examples.

Freshness

Last updated 2026-02-06 03:58:06.