bge-small-en-v1.5

bge-small-en-v1.5

A small English text embedding model from BAAI that turns sentences into vectors for search and retrieval. About 33M parameters, and part of the FlagEmbedding project.

Made by BAAI on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: bge-small-en-v1.5 is a small English embedding model from BAAI. It turns a sentence or passage into a vector of numbers, so that texts with similar meaning end up close together. It has 33.4M parameters and belongs to the FlagEmbedding project, which centers on retrieval for language models.

Who it suits: Developers who build search, question answering or retrieval-augmented apps in English and want a light model. Its small size makes it easy to run on modest hardware.

Industry | Who uses it | How they use it | Example
Software | App developers | Add semantic search to a product | Embed help articles and match them to a user question
AI applications | Builders of retrieval-augmented tools | Fetch relevant passages for a language model | Embed a document set and look up the closest chunks
Data work | Analysts | Group similar texts together | Cluster support tickets by meaning
Research | NLP researchers | Fine-tune or compare embedding models | Start from this small model and train it on your own pairs

You can skip it if you need other languages or long texts. The card points to bge-m3 for more languages, longer inputs and other retrieval methods.

Where to use it:

• Python on Windows, macOS and Linux, through the sentence-transformers library
• Retrieval pipelines that compare vectors
• Re-ranking setups, where the card suggests its reranker models for the top results

Use case | Who | What they get back
Semantic search | Developers | Vectors you compare to find the closest texts
Retrieval for language models | AI builders | The passages most similar to a question
Clustering | Analysts | Groups of texts with similar meaning

How to set it up: You need Python and a small amount of disk space.

• Python on Windows, macOS or Linux
• The sentence-transformers library
• About 133 MB for the F32 weights

1. Install Python if you do not have it.
2. Install the Hugging Face hub tool.
3. Download the model files, or let your library fetch them by name.

pip install -U huggingface_hub
hf download BAAI/bge-small-en-v1.5

How to use it: The part of the card shown here has no code. In practice you load the model by name through sentence-transformers, encode your texts into vectors, and compare the vectors to rank them by similarity.

Example: You encode a user question and a set of help articles, then compare the vectors. The article closest to the question comes back first.

For newcomers and experts: An embedding is a list of numbers that stands for the meaning of a text. Comparing two lists tells you how alike the texts are. A first small step is to embed two sentences you know are related and one that is not, and see which pair scores closest.

Version 1.5 evens out the spread of similarity scores and improves retrieval when you give no instruction. The card says the small and base sizes released in 2023 performed best among models of their size. For better ranking, the card suggests running a bge reranker over the top results from an embedding model.

Load the model by name
Encode your texts into vectors
Compare vectors for similarity
Rank or retrieve the closest texts

Good to know:

• It is an English model with 33.4M parameters.
• Other people have published 41 quantized and 555 fine-tuned versions.
• The card lists FlagEmbedding's other parts, including rerankers, long-context work and a benchmark.
• License: MIT.
• Cost to run: none named.
• Data it reaches: the texts you embed, plus files downloaded from Hugging Face.
• Not for: languages other than English, where the card suggests bge-m3.

On Hugging Face, 13 November 2026

• Task: feature extraction
• Library: sentence-transformers
• Parameters: 33.4M
• Weights: about 133 MB in F32; about 17 MB at 4-bit (estimate)
• Formats: Safetensors
• Downloads in the last 30 days: 45,750,548
• Likes: 671
• License: mit
• Languages: en
• Quantized versions: 42
• Fine-tunes: 556
• Tags: sentence-transformers, pytorch, onnx, safetensors, bert, feature-extraction, sentence-similarity, transformers, mteb, model-index, text-embeddings-inference

Links

• Model: https://huggingface.co/BAAI/bge-small-en-v1.5
• Files: https://huggingface.co/BAAI/bge-small-en-v1.5/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download BAAI/bge-small-en-v1.5

Then follow the usage in its model card. Nothing in this kit runs by itself.