# FlagOpen/FlagEmbedding

Retrieval and Retrieval-augmented LLMs

Repository: https://github.com/FlagOpen/FlagEmbedding
Canonical: https://ross.abutalabs.com/products/flagembedding
Homepage: http://www.bge-model.com/
Language: Python
License: MIT
License Family: permissive
Topics: embeddings, information-retrieval, llm, sentence-embeddings, text-semantic-similarity, retrieval-augmented-generation
Last push: 2026-08-24T02:57:42+00:00

## Health v2 (maintenance only)
Score: 87/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 75, longevity 80
- inputs: {"age_days": 1128, "days_push": 9, "days_rel": 9, "gap_med": 109, "n_releases_24m": 6}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 12086, forks 908 (observed 2026-08-28T04:10:51.727306+00:00)

## What it is
FlagEmbedding is the official Python toolkit for BAAI's BGE family of embedding models and rerankers, covering inference, evaluation, and fine-tuning for search and retrieval-augmented generation. It includes dense, sparse, and multi-vector retrieval models (e.g., bge-m3) plus multimodal embedding models (BGE-VL).

## Use cases
- generate sentence embeddings for semantic search
- build a RAG pipeline with retrieval
- rerank retrieved documents for better relevance
- fine-tune a custom embedding model on my own data
- compute text semantic similarity between sentences
- multilingual text retrieval across 100+ languages
- multimodal image and text search with embeddings

## When to choose
- you need state-of-the-art open embedding or reranker models for search or RAG
- you want dense, sparse, and multi-vector retrieval in one toolkit
- you need multilingual embeddings or fine-tuning support
- you want MIT-licensed models free for commercial use

## When to avoid
- you need a lightweight general-purpose sentence-transformers wrapper without BGE-specific features
- you want a hosted embedding API rather than running models locally
- you lack GPU resources for large-scale encoding or fine-tuning

## Facets
- artifact type: library
- maturity: active
- function: rag, search-engine, machine-learning, nlp, llm-inference
- domain: machine-learning, large-language-models
- platform: python, cross-platform
- tags: embeddings, sentence-transformers, reranker, bge, semantic-search, multilingual, dense-retrieval, fine-tuning, retrieval-augmented-generation, natural-language-processing, search, gpu

## Member repositories
- FlagOpen/FlagEmbedding (main) score 87

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:51.727306+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:14:56.868160+00:00, confidence not recorded.
  - readme: https://github.com/FlagOpen/FlagEmbedding (fetched 2026-08-28T04:10:51.727306+00:00, sha 193b9d8c4d6a)
  - homepage: http://www.bge-model.com/ (fetched 2026-08-29T08:12:17.996781+00:00, sha dc1da902e787)
  - registry_pypi: https://pypi.org/pypi/flagembedding/json (fetched 2026-08-29T08:12:18.008143+00:00, sha 822840579777)
  - site_page: https://bge-model.com/FAQ/index.html (fetched 2026-08-29T08:12:18.006279+00:00, sha b426473f6041)
- Data as of 2026-08-30T08:39:29.467469+00:00.