embeddings-benchmark/mteb
A Python benchmark suite for evaluating text embedding and retrieval models across multiple NLP tasks.
Not currently ranked — collecting fresh signals.
star history
MTEB (Massive Text Embedding Benchmark) is a toolbox for evaluating embedding and retrieval systems. It provides standardized benchmarks across tasks including semantic search, clustering, reranking, STS, bitext mining, and text classification. The repository offers a leaderboard on HuggingFace for comparing model performance and supports installation via pip or uv.
Frequently asked
- What is embeddings-benchmark/mteb?
- A Python benchmark suite for evaluating text embedding and retrieval models across multiple NLP tasks.
- Is mteb open source?
- Yes — embeddings-benchmark/mteb is open source, released under the Apache-2.0 license.
- What language is mteb written in?
- embeddings-benchmark/mteb is primarily written in Python.
- How popular is mteb?
- embeddings-benchmark/mteb has 3.4k stars on GitHub.
- Where can I find mteb?
- embeddings-benchmark/mteb is on GitHub at https://github.com/embeddings-benchmark/mteb.