Open VLM Leaderboard

Open VLM Leaderboard

Shows leaderboards of vision-language model scores gathered with the VLMEvalKit toolkit, with an overall ranking and separate tables for each benchmark. It suits researchers and engineers comparing multimodal models.

Made by opencompass 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: This Space is a leaderboard of scores for vision-language models, all measured by one open-source evaluation toolkit called VLMEvalKit. It has an overall board built from a hand-picked set of benchmarks, plus a board for each separate benchmark with total and finer scores. It runs on cpu-upgrade hardware at Hugging Face.

Who it suits: It suits people who need to compare vision-language models and want benchmark scores in one place.

Industry | Who uses it | How they use it | Example
AI research | Researchers | Compare model scores across benchmarks | Check the overall ranking, then open one benchmark for details
AI engineering | ML engineers | Shortlist a vision-language model for a product | Look at fine-grained scores for the benchmark closest to their task
Education | Students | Learn which benchmarks exist for multimodal models | Browse the benchmark-level tables

You can skip it if you want to run an evaluation yourself, since the page shows results and does not run tests.

Where to use it: You can use it in these places:

• A browser: open the Space's page on Hugging Face and nothing needs installing.
• Windows, macOS or Linux: run your own copy with Python 3.10.20.

Use case | Who | What they get back
Overall ranking | Researchers | A score table over a curated set of benchmarks
Benchmark detail | Engineers | Overall and fine-grained scores for one benchmark

How to set it up: You need these first for a local copy:

• Python 3.10.20 and pip
• A terminal to run the commands

The online page needs none of this. For your own copy, follow the steps below.

1. Install the Hugging Face tools.
2. Download the Space's files into a folder.
3. Move into that folder.
4. Install Gradio at the version the Space uses, then the other packages.
5. Start the app.

pip install -U huggingface_hub
hf download opencompass/open_vlm_leaderboard --repo-type space --local-dir open_vlm_leaderboard
cd open_vlm_leaderboard
pip install gradio==4.44.1
pip install -r requirements.txt
python app.py

How to use it: Open the Space in a browser, or start your own copy with the command below. Then read the overall leaderboard or open the table for one benchmark.

python app.py

Tool | What it does
Overall leaderboard | Ranks models on a curated selection of benchmarks with an overall score
Benchmark leaderboards | One table per benchmark, with the total score and finer detail

Example: You open the overall leaderboard to see how the models rank. Then you open the table for one benchmark to read the more detailed scores for each model.

For newcomers and experts: If you are new to this, start with the overall leaderboard and treat it as a quick guide. Then open the benchmark that is closest to your own use.

If you know the field, all numbers come from VLMEvalKit runs, so the scores share one evaluation toolkit. The repository is small, with app.py building the page and gen_table.py and meta_data.py supplying the tables and model information. The page asks readers to cite the VLMEvalKit report when it helps their research.

VLMEvalKit evaluates models
Results are collected here
Overall table is built
Benchmark tables are built
You compare the scores

Good to know: The Space shows results and does not run evaluations itself.

• License: apache-2
• Cost to run: none named
• Data it reaches: none named, since it only shows collected results
• Not for: running evaluations, which are done with VLMEvalKit

On Hugging Face, 13 November 2026

• Likes: 1,034
• Built with: gradio
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/opencompass/open_vlm_leaderboard
• Files: https://huggingface.co/spaces/opencompass/open_vlm_leaderboard/tree/main

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.