[Models](/models)/[Embeddings](/models?category=embeddings)/BAAI/bge-m3

HFText Embeddingsmit

# bge-m3

by BAAI

Hybrid retrieval in one model -- dense, sparse, and ColBERT embeddings from a single forward pass

36.1Mdl/month

3,439likes

568Mparams

[HuggingFace](https://huggingface.co/BAAI/bge-m3)[Run it on your own data, free](https://studio.mixpeek.com/namespaces/create?mode=managed&utm_source=homepage&utm_medium=model_page&utm_campaign=mvs_onramp)

Identifiers

Model ID

`BAAI/bge-m3`

## Deploy bge-m3

Single-tenant

Mixpeek has no managed extractor for this model. On a single-tenant deployment you upload the weights and a custom plugin serves them next to the rest of your pipeline.

[Read the customer-hosted guide](https://mixpeek.com/docs/customer-hosted/overview) [Talk to us about single-tenant](/demo)

## Overview

BGE-M3 is BAAI's multi-functionality embedding model that produces three types of embeddings simultaneously: dense vectors for semantic search, sparse vectors for lexical matching, and ColBERT-style multi-vector representations for fine-grained late interaction. This eliminates the need to run separate models for different retrieval strategies.

The model supports 100+ languages and handles up to 8192 tokens of input, making it suitable for long documents. On Mixpeek, BGE-M3 powers hybrid retrieval pipelines where a single ingest pass produces all three representation types, and the retriever fuses them at query time for higher recall than any single strategy alone.

## Architecture

XLM-RoBERTa backbone, 568M parameters. Produces dense embeddings (1024d), sparse term-weight vectors, and ColBERT multi-vector representations from one forward pass. Trained with self-knowledge distillation across 100+ languages.

## Mixpeek SDK Integration

IngestRetrieve

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so bge-m3 runs
// on your side and the output is upserted through POST
// /v1/namespaces/{namespace\_id}/documents/upsert. On Enterprise the other
// path is to upload the weights instead: POST /v1/namespaces/{id}/models
// accepts the huggingface format and a custom plugin loads them.
const res = await fetch(
  "https://api.mixpeek.com/v1/namespaces/ns\_your\_namespace/documents/upsert",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer API\_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      collection\_id: "col\_your\_collection",
      documents: \[
        {
          document\_id: "asset-00412",
          // The vector name has to match a vector index on the collection.
          vectors: { "text-embedding": yourVector },
          payload: { source\_key: "archive/2026/asset-00412" },
        },
      \],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// text\_extractor@v1 runs intfloat/multilingual-e5-large-instruct
// (1024-d) over a bucket, with no inference of your own.

## Capabilities

*   Dense, sparse, and ColBERT embeddings in one pass
*   100+ language support
*   8192 token context window
*   Hybrid retrieval without multiple models
*   Matryoshka dimension reduction

## Use Cases on Mixpeek

Hybrid search combining semantic + lexical matching

Cross-lingual document retrieval

Long-document embedding for RAG pipelines

Multi-strategy retrieval with single model overhead

## Benchmarks

Dataset

Metric

Score

Source

MIRACL (avg, 18 languages)

nDCG@10

71.9%

BAAI, 2024 -- Paper Table 3

MTEB Retrieval (en)

nDCG@10

67.2%

MTEB Leaderboard

## Performance

Input SizeUp to 8192 tokens

GPU Latency~12ms / passage (A100)

GPU Throughput~800 passages/sec (A100, batch 64)

GPU Memory~2.2 GB

## Common Pipeline Companions

[

Qwen/Qwen3-Reranker-8B

Rerank BGE-M3 hybrid results for precision

](/model/Qwen/Qwen3-Reranker-8B)[

BAAI/bge-large-en-v1.5

Compare dense-only vs hybrid retrieval

](/model/BAAI/bge-large-en-v1.5)

## Explore on Mixpeek

[

More Embeddings models

Compare alternatives in this category

](/models?category=embeddings)[

Best Self-Hosted Embedding Models

Hand-picked tools & platforms compared

](/curated-lists/best-self-hosted-embedding-models)[

Instruction-Tuned Embeddings

Deep-dive technical guide

](/guides/instruction-tuned-embeddings)[

Feature Extractors

See how Mixpeek runs models as extractors

](/extractors)[

Mixpeek Vector Store

Store & search embeddings at scale

](/mvs)[

Pricing

Usage-based pricing for pipelines

](/pricing)[

All Curated Lists

Compare models, APIs & infrastructure

](/curated-lists)

### Specification

FrameworkHF

OrganizationBAAI

FeatureText Embeddings

Output1024-dim vector

Modalitiesdocument, audio

RetrieverText Similarity

Parameters568M

Licensemit

Downloads/mo36.1M

Likes3,439

[

### Research Paper

BGE M3-Embedding: Multi-Lingual, Multi-Functionality, Multi-Granularity

arxiv.org](https://arxiv.org/abs/2402.03216)

### Build a pipeline with bge-m3

Add this model to a processing pipeline alongside other extractors. Combine with retrieval stages for end-to-end search.

[Run it on your own data, free](https://studio.mixpeek.com/namespaces/create?mode=managed&utm_source=homepage&utm_medium=model_page&utm_campaign=mvs_onramp)

### Alternative Models

[

BAAI/bge-large-en-v1.5

Text Embeddings

](/model/BAAI/bge-large-en-v1.5)[

sentence-transformers/all-MiniLM-L6-v2

Text Embeddings

](/model/sentence-transformers/all-MiniLM-L6-v2)[

Qwen/Qwen3-Embedding-0.6B

Text Embeddings

](/model/Qwen/Qwen3-Embedding-0.6B)[

nomic-ai/nomic-embed-text-v2-moe

Text Embeddings

](/model/nomic-ai/nomic-embed-text-v2-moe)

### Related in Embeddings

[

openai/clip-vit-large-patch14

Visual Embeddings

](/model/openai/clip-vit-large-patch14)[

google/siglip-base-patch16-224

Visual Embeddings

](/model/google/siglip-base-patch16-224)[

google/siglip2-giant-opt-patch16-384

Visual Embeddings

](/model/google/siglip2-giant-opt-patch16-384)[

facebook/dinov2-large

Visual Embeddings

](/model/facebook/dinov2-large)