Mindket
English
English简体中文
Sign inSign up
StoreMCPSkillsCalculatorLearnPriceBlog
Discover
Research312Writing59Business95Legal13Finance & accounting33Marketing & SEO175Developer1,422Productivity391Education81Companions14
Help and support
Language
English简体中文

MTEB Leaderboard

Made by mtebKitResearch
Newversion 1 · updated
Like

Overview

Ranks embedding models for text, images, audio and video on a leaderboard page that runs in the browser or in Docker. For people choosing which embedding model to use.

MTEB 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
  • Python and pip, to fetch the files
  • Windows, macOS or Linux
Runs in
  • A web browser, on the Space's page at Hugging Face
  • Docker on your own Windows, macOS or Linux computer
Start with
pip install -U huggingface_hub
hf download mteb/leaderboard --repo-type space --local-dir leaderboard
cd leaderboard

Platform

DockerContainer
WindowsOperating system
macOSOperating system
LinuxOperating system

About

What it is

MTEB Leaderboard is a web page for comparing embedding models. It covers models for text, images, audio and video. You can open it online, or build and run your own copy with Docker.

Who it suits

It suits people who need to pick an embedding model and want a ranking to start from, and developers curious how such a leaderboard is hosted.

IndustryWho uses itHow they use itExample
AI researchResearchersCompare embedding models side by sideLook up a model you already follow
SoftwareML engineersShortlist a model for their own projectCompare a few candidates before testing them
Multimodal AIDevelopers working with images, audio or videoFind models beyond textBrowse the models listed for another media type
OperationsPlatform engineersStudy how the leaderboard is containerisedRead the Dockerfile and models.py

Skip it if you need to know how each score is produced, since the source gives no method.

Where to use it

  • A web browser, on the Space's page at Hugging Face
  • Docker on your own Windows, macOS or Linux computer
Use caseWhoWhat they get back
Find an embedding modelEngineersA ranked view of models to start from
Compare across mediaDevelopersEntries for text, image, audio or video models
Run a private copyPlatform teamsThe same page running from a container on your machine

How to set it up

The online page needs nothing. A local copy needs Docker, and Python with pip to fetch the files.

  • Docker
  • Python and pip, to fetch the files
  • Windows, macOS or Linux
  1. Install the Hugging Face hub library.
  2. Download the Space's files into a folder named leaderboard.
  3. Move into that folder.
  4. Build the Docker image.
  5. Run the container with port 7860 published, then open that port in your browser.
pip install -U huggingface_hub
hf download mteb/leaderboard --repo-type space --local-dir leaderboard
cd leaderboard
docker build -t leaderboard .
docker run -it -p 7860:7860 leaderboard

How to use it

Open the Space's page, or start the container with the last command above and browse to port 7860 on your machine. Then look through the leaderboard for the media type you care about.

docker run -it -p 7860:7860 leaderboard

Example

You open the page and look for models that handle text. You compare where a few of them sit before deciding which ones to try in your own project.

For newcomers and experts

Open the online page and scroll through it. An embedding model turns content into numbers so that similar items sit close together, and the leaderboard helps you pick one.

The app is built with Docker, serves on port 7860 and has app.py as its main file. The Hugging Face copy runs on the cpu-upgrade tier. The repository itself holds only a Dockerfile, a README and models.py, and the source does not say where the page's data comes from. The record names many models, among them German_Semantic_STS_V2, MedEmbed-small-v0.1 and FRIDA, and one dataset called BioVITA.

  1. Docker builds the image
  2. The container runs app.py
  3. The page is served on port 7860
  4. You browse the leaderboard

Good to know

  • Hardware: The Hugging Face copy runs on the cpu-upgrade tier.
  • Gaps: The source says little beyond its purpose, so scoring methods and columns are not described here.
  • License: mit.
  • Cost to run: none named.
  • Data it reaches: The source does not say what data the page reads.

README and files

README

MTEB Leaderboard

Ranks embedding models for text, images, audio and video on a leaderboard page that runs in the browser or in Docker. For people choosing which embedding model to use.

Made by mteb 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: MTEB Leaderboard is a web page for comparing embedding models. It covers models for text, images, audio and video. You can open it online, or build and run your own copy with Docker.

Who it suits: It suits people who need to pick an embedding model and want a ranking to start from, and developers curious how such a leaderboard is hosted.

Industry | Who uses it | How they use it | Example
AI research | Researchers | Compare embedding models side by side | Look up a model you already follow
Software | ML engineers | Shortlist a model for their own project | Compare a few candidates before testing them
Multimodal AI | Developers working with images, audio or video | Find models beyond text | Browse the models listed for another media type
Operations | Platform engineers | Study how the leaderboard is containerised | Read the Dockerfile and models.py

Skip it if you need to know how each score is produced, since the source gives no method.

Where to use it:

  • A web browser, on the Space's page at Hugging Face
  • Docker on your own Windows, macOS or Linux computer
Use case | Who | What they get back
Find an embedding model | Engineers | A ranked view of models to start from
Compare across media | Developers | Entries for text, image, audio or video models
Run a private copy | Platform teams | The same page running from a container on your machine

How to set it up: The online page needs nothing. A local copy needs Docker, and Python with pip to fetch the files.

  • Docker
  • Python and pip, to fetch the files
  • Windows, macOS or Linux
  1. Install the Hugging Face hub library.
  2. Download the Space's files into a folder named leaderboard.
  3. Move into that folder.
  4. Build the Docker image.
  5. Run the container with port 7860 published, then open that port in your browser.
pip install -U huggingface_hub
hf download mteb/leaderboard --repo-type space --local-dir leaderboard
cd leaderboard
docker build -t leaderboard .
docker run -it -p 7860:7860 leaderboard

How to use it: Open the Space's page, or start the container with the last command above and browse to port 7860 on your machine. Then look through the leaderboard for the media type you care about.

docker run -it -p 7860:7860 leaderboard

Example: You open the page and look for models that handle text. You compare where a few of them sit before deciding which ones to try in your own project.

For newcomers and experts: Open the online page and scroll through it. An embedding model turns content into numbers so that similar items sit close together, and the leaderboard helps you pick one.

The app is built with Docker, serves on port 7860 and has app.py as its main file. The Hugging Face copy runs on the cpu-upgrade tier. The repository itself holds only a Dockerfile, a README and models.py, and the source does not say where the page's data comes from. The record names many models, among them German_Semantic_STS_V2, MedEmbed-small-v0.1 and FRIDA, and one dataset called BioVITA.

Docker builds the image
The container runs app.py
The page is served on port 7860
You browse the leaderboard

Good to know:

  • Hardware: The Hugging Face copy runs on the cpu-upgrade tier.
  • Gaps: The source says little beyond its purpose, so scoring methods and columns are not described here.
  • License: mit.
  • Cost to run: none named.
  • Data it reaches: The source does not say what data the page reads.

On Hugging Face, 13 November 2026

  • Likes: 7,773
  • Built with: docker
  • License: mit

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 · 2026-11-12

Compare with similar

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

ProductLikesBuilt withLicenseLast change
MTEB Leaderboard This page7,773dockermit
Open LLM LeaderboardShares docker · leaderboard14,156dockerapache-2.0
UGI LeaderboardShares docker · leaderboard2,199dockerapache-2.0
CADGenBench LeaderboardShares docker · leaderboard90dockerapache-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.

davidtsongShow HN: Playground for comparing embedding models on Wikipedia+book retrievalIntroducing embeds.ai: an embedding playground to compare how embedding models work on a real world use case (retrieval augmented generation for Wikipedia articles + Elad Gil's High growth handbook) A few weeks ago, Shreyan and I were looking for an embedding model to use for RAG. We eventually…5 points · 11 repliesMTEB1/ The best embedding model you can afford to run can sit well down the MTEB leaderboard, because the ranking ignores size. A new Pareto tag now marks it: no model with the same or fewer active parameters has a higher average score. Hover it to see the size range it leads.2 likes · 1 replylevkkShow HN: We unified LLMs, vector memory, ranking, pruning models in one processThere is a lot of latency involved shuffling data for modern and complex ML systems in production. In our experience these costs dominate end-to-end user latency, rather than actual model or ANN algorithms, which unfortunately limits what is achievable for interactive applications. We've extended…4 points · 1 replyMTEB3/ Because scores change, we will rerun the audio and video models on the leaderboard with the new settings. If you submit new audio or video results, please use mteb 2.24.4 or later. Work by AdnanElAssadi.2 likesyokeeShow HN: SOTA Sentence Embeddings on MTEB Leaderboard3 points@qdrddr.bsky.socialHope to see its performance on MTEB leaderboard soon.yokeeUniversal AnglE Sentence Embedding: New SOTA on MTEB Leaderboard2 points · 1 replyGunter RotsaertThe Hugging Face massive text embedding benchmark leaderboard.caprockText Embedding Benchmark (MTEB) Leaderboard1 point · 1 replySurajSILT#00395: If you're looking for embedding models, consider looking at the MTEB Embedding Leaderboard on HuggingFace to see a wide selection of models across many benchmarks, as well as compare them with each other to see a good fit for your task: huggingface.co/spaces/mteb/...t4Ask HN: Help for Choosing an Embedding Model?I've tried looking and the MTEB leaderboard and tested a few, but I'm curious about your experiences! I'm looking for a model that would optimize for accuracy (finding relevant results), language support (as many as possible, English only is a no-no) and efficiency so I could potentially run it…1 pointGunter RotsaertThe Hugging Face massive text embedding benchmark leaderboard.

Searched

More from Mindket

See all from Mindket

anaksor-ocr

Sokheng · Kit

Reads Khmer and English text from documents, receipts and photos of signs, using a detector for text lines and a recogniser for each line. Suits developers who…

New

Qwen2.5-7B-Instruct-GGUF by Qwen

Qwen · Kit

Qwen2.5-7B-Instruct as GGUF files in eight quantization sizes, to chat with locally through llama.cpp. For people who want a capable 7.6B model on their own co…

New

ACE-Step 1.5 XL SFT

ACE-Step · Kit

Generates music from text with a 4B diffusion decoder, 50 steps and guidance control, aiming for higher audio quality than the base. For creators with a GPU of…

New

Similar in Research

See all in Research

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

getting-over-it-rl-jax

zimmerk4 · Kit

Plays the full game Getting Over It from the bottom to the end with a small trained policy network that learned inside a JAX physics simulator. For reinforceme…

New