Mindket

CADGenBench Leaderboard

Newversion 1 · updated

Overview

Ranks AI systems that turn a text or image description of a mechanical part into a valid 3D CAD model, and scores uploaded STEP or BREP files. For CAD generation researchers and engineers.

CADGenBench Leaderboard is a kit in the Research category, listed by Mindket. It is free. This is version 1, last updated on . Download it on this page, a zip of its files. Nothing in it runs by itself.

At a glance

You need
  • Docker on your computer
  • The huggingface_hub package, which gives you the hf command
Runs in
  • A web browser, on the Space's page at Hugging Face, with nothing to install
  • Windows, macOS or Linux with Docker, from a copy you download
Start with
pip install -U huggingface_hub
hf download HuggingAI4Engineering/CADGenBench --repo-type space --local-dir CADGenBench
cd CADGenBench

Platform

WindowsOperating system
macOSOperating system
LinuxOperating system
DockerContainer

About

What it is

CADGenBench Leaderboard is the results page for a benchmark of AI-driven CAD generation. It measures whether a system can read a written or pictured description of a machine part and return a 3D model that is valid and geometrically right.

The Space has three tabs. Leaderboard shows ranked submissions, Submit scores your own files, and About explains the scoring axes.

Who it suits

Researchers and engineers who build or compare systems that generate CAD models.

IndustryWho uses itHow they use itExample
Mechanical engineeringCAD engineersSee how generation systems rankOpen the Leaderboard tab and read the rows
ResearchAI researchersScore their own model on the benchmarkUpload STEP files on the Submit tab
SoftwareTool buildersCompare approaches to CAD generationRead a submission's report

You can skip it if you do not work with CAD generation, since the Space only covers this benchmark.

Where to use it

  • A web browser, on the Space's page at Hugging Face, with nothing to install
  • Windows, macOS or Linux with Docker, from a copy you download
Use caseWhoWhat they get back
Browse rankingsResearchersA ranked list of submissions
Submit candidatesModel buildersA scored row with a per-submission report
Read the scoringAnyoneAn explanation of the scoring axes

How to set it up

  • Docker on your computer
  • The huggingface_hub package, which gives you the hf command
  1. Install the huggingface_hub package.
  2. Download the Space with the hf command.
  3. Move into the downloaded folder.
  4. Build the Docker image.
  5. Run the image with port 7860 open.
pip install -U huggingface_hub
hf download HuggingAI4Engineering/CADGenBench --repo-type space --local-dir CADGenBench
cd CADGenBench
docker build -t cadgenbench .
docker run -it -p 7860:7860 cadgenbench

How to use it

Open the Space's page in your browser. Read the Leaderboard tab for rankings. To submit, go to the Submit tab, upload your STEP or BREP candidate files and wait for the Space to run the evaluation.

ToolWhat it does
Leaderboard tabShows ranked submissions
Submit tabTakes STEP or BREP files and runs the evaluation
About tabExplains the scoring axes and where the benchmark code lives

Example

You open the Submit tab and upload STEP files your system produced. The Space runs the evaluation and adds a row for your submission, with a report you can open.

For newcomers and experts

If you are new to CAD generation, start on the Leaderboard tab and read the top rows. Then open the About tab to see how scoring works.

If you know the field, the Space is a Docker app that serves several pages, including leaderboard, submit, gallery, metrics and admin views. It names two datasets, one for submissions and one for benchmark data. The repository also holds a tests folder and a tools folder.

  1. You upload STEP or BREP files
  2. The Space runs the evaluation
  3. A row is published with a report
  4. The row joins the leaderboard

Good to know

  • License: the Space lists apache-2
  • Cost to run: none named
  • Data it reaches: the files you upload, plus the two datasets it names
  • Not for: scoring anything other than STEP or BREP candidate files

README and files

README

CADGenBench Leaderboard

Ranks AI systems that turn a text or image description of a mechanical part into a valid 3D CAD model, and scores uploaded STEP or BREP files. For CAD generation researchers and engineers.

Made by HuggingAI4Engineering on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: CADGenBench Leaderboard is the results page for a benchmark of AI-driven CAD generation. It measures whether a system can read a written or pictured description of a machine part and return a 3D model that is valid and geometrically right.

The Space has three tabs. Leaderboard shows ranked submissions, Submit scores your own files, and About explains the scoring axes.

Who it suits: Researchers and engineers who build or compare systems that generate CAD models.

Industry | Who uses it | How they use it | Example
Mechanical engineering | CAD engineers | See how generation systems rank | Open the Leaderboard tab and read the rows
Research | AI researchers | Score their own model on the benchmark | Upload STEP files on the Submit tab
Software | Tool builders | Compare approaches to CAD generation | Read a submission's report

You can skip it if you do not work with CAD generation, since the Space only covers this benchmark.

Where to use it:

  • A web browser, on the Space's page at Hugging Face, with nothing to install
  • Windows, macOS or Linux with Docker, from a copy you download
Use case | Who | What they get back
Browse rankings | Researchers | A ranked list of submissions
Submit candidates | Model builders | A scored row with a per-submission report
Read the scoring | Anyone | An explanation of the scoring axes

How to set it up:

  • Docker on your computer
  • The huggingface_hub package, which gives you the hf command
  1. Install the huggingface_hub package.
  2. Download the Space with the hf command.
  3. Move into the downloaded folder.
  4. Build the Docker image.
  5. Run the image with port 7860 open.
pip install -U huggingface_hub
hf download HuggingAI4Engineering/CADGenBench --repo-type space --local-dir CADGenBench
cd CADGenBench
docker build -t cadgenbench .
docker run -it -p 7860:7860 cadgenbench

How to use it: Open the Space's page in your browser. Read the Leaderboard tab for rankings. To submit, go to the Submit tab, upload your STEP or BREP candidate files and wait for the Space to run the evaluation.

Tool | What it does
Leaderboard tab | Shows ranked submissions
Submit tab | Takes STEP or BREP files and runs the evaluation
About tab | Explains the scoring axes and where the benchmark code lives

Example: You open the Submit tab and upload STEP files your system produced. The Space runs the evaluation and adds a row for your submission, with a report you can open.

For newcomers and experts: If you are new to CAD generation, start on the Leaderboard tab and read the top rows. Then open the About tab to see how scoring works.

If you know the field, the Space is a Docker app that serves several pages, including leaderboard, submit, gallery, metrics and admin views. It names two datasets, one for submissions and one for benchmark data. The repository also holds a tests folder and a tools folder.

You upload STEP or BREP files
The Space runs the evaluation
A row is published with a report
The row joins the leaderboard

Good to know:

  • License: the Space lists apache-2
  • Cost to run: none named
  • Data it reaches: the files you upload, plus the two datasets it names
  • Not for: scoring anything other than STEP or BREP candidate files

On Hugging Face, 13 November 2026

  • Likes: 90
  • Built with: docker
  • License: apache-2.0

Links

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.

Files
README.md  4 KB

1 file, 4 KB, downloaded as a zip.

What’s new

  • Version 1 · 2026-11-13

Compare with similar

Listed Hugging Face Spaces in the same category that share the most tags with this one.

ProductLikesBuilt withLicenseLast change
CADGenBench Leaderboard This page90dockerapache-2.0
Open LLM LeaderboardShares docker · leaderboard14,156dockerapache-2.0
MTEB LeaderboardShares docker · leaderboard7,773dockermit
UGI LeaderboardShares docker · leaderboard2,199dockerapache-2.0
Microduck SandboxShares docker804dockerapache-2.0
Visualize Dataset (v2.0+ latest dataset format)Shares docker754dockerapache-2.0
Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13.

Reviews

No ratings yet

Write a review

No reviews yet.

What people say

Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search and on Trendshift. Mindket doesn’t write or check them. Open one to read it where it was posted.

Searched

More from Mindket

See all from Mindket

DeepSeek-R1-GGUF

unsloth · Kit

GGUF copies of the 671B DeepSeek-R1 reasoning model, with 1.58-bit and 2-bit dynamic quants, for running it in llama.cpp on a local machine with enough memory.

New

Axion-1.0-35B-GGUF

prithivMLmods · Kit

Provides GGUF files of Axion-1.0-35B, a 35B mixture-of-experts model with about 3B active parameters aimed at scientific reasoning, information seeking and too…

New

HEART

Phips · Kit

Upscales low-resolution images with a super-resolution network that trains on one consumer GPU and exports to ONNX. For people who train or run image upscalers.

New

Similar in Research

See all in Research

ManiFlow-tb

allenai · Kit

The low-level action model of allenai's ToyBlocks block-stacking project, trained on simulated Franka arm data to pick, place and rotate three blocks on a tabl…

New

Merlin-Cancer-Super

AbdomenAtlas · Kit

Detects and outlines tumors in 9 organs on CT scans, trained with the R-Super method that learns from radiology reports. A research model for medical imaging g…

New

EarthJev

dBenf · Kit

Answers typed questions about satellite image patches from all 12 Sentinel-2 bands, returning a probability for each option without writing text. A Qwen3.5-0.8…

New