skip to content
gigarouter gigarouter

BGE Small EN v1.5

BAAI/bge-small-en-v1.5

published Sep 2023 · updated Feb 2024

BGE Small EN v1.5 is a dense embedding model that converts English text into high-quality vector representations for semantic search and retrieval.

price
$0.008
/ 1M tokens
API providers
0
throughput
193 embeds/s
license
mit

specs

TaskEmbedding
ArchitectureTransformer encoder
ParametersSmall (approx. 33M)
LicenseMIT

about this model

BAAI/bge-small-en-v1.5 is an English text embedding model that generates dense vector representations optimized for semantic similarity and retrieval tasks. Developed as part of the BGE (BAAI General Embedding) series, this v1.5 release delivers a more reasonable similarity distribution compared to its predecessor, improving the reliability of cosine similarity scores for ranking and search.

The model is designed for efficiency: it is a small-scale model that retains competitive performance relative to larger alternatives, making it suitable for latency-sensitive or resource-constrained deployments. For retrieval use cases, the recommended practice is to prepend the query with the instruction "Represent this sentence for searching relevant passages: " (no instruction is added to passages). This aligns with the broader BGE family’s approach to separating query and document encoding.

BAAI/bge-small-en-v1.5 is hosted on gigarouter as a managed API, eliminating the need for local installation or manual inference setup. Developers can integrate it via OpenAI-compatible endpoints for embedding generation, benefiting from the model’s proven performance without infrastructure overhead.

best for

FAQ

What is BGE Small EN v1.5 best for?

It is best for English text embedding tasks like semantic search, retrieval-augmented generation, and sentence similarity where low latency and small model size are important.

How does it compare in size and speed to larger BGE models?

BGE Small EN v1.5 has fewer parameters (about 33M) than base and large versions, making it faster and more memory-efficient, though with slightly lower accuracy on benchmarks.

What are the input and output formats for the API?

The API accepts text strings as input and returns a vector (list of floats) as output. Use the gigarouter OpenAI-compatible endpoint with an API key.

Does it require an instruction prefix for queries?

Yes, for retrieval tasks you should prepend the query with "Represent this sentence for searching relevant passages: ". No instruction is needed for documents.

What is the license for using this model?

The model is released under the MIT license, allowing free use, modification, and distribution.

call it
# OpenAI client - just change base_url
from openai import OpenAI
client = OpenAI(base_url="https://gigarouter.ai/v1", api_key=KEY)
v = client.embeddings.create(model="BAAI/bge-small-en-v1.5", input=["hello world"])
print(v.data[0].embedding[:4])

try it live

runs the real hosted model on a shared demo allowance · get your own key + $25 free →

related embeddings models

compare all →