Usage-based · Free plan🇨🇦
Embedding model from Canadian vendor Cohere for text and images in 100+ languages, via API, dedicated or private deployment.
BGE Reranker v2 M3 overview and key facts (October 2026)
Open multilingual reranker model from the Beijing Academy of Artificial Intelligence under Apache 2.0, light enough for your own servers.
Free
What is BGE Reranker v2 M3?
BGE Reranker v2 M3 is an open-weight model that outputs a relevance score for each pair of query and passage. Developers use it to reorder search results in RAG systems they host themselves.
The model comes from the Beijing Academy of Artificial Intelligence (BAAI), a research organization that also publishes the BGE embedding models and the FlagEmbedding library. It is built on bge-m3, supports multiple languages and is released under the Apache 2.0 license. With about 17 million downloads on Hugging Face in the last month as of September 28, 2026, it is among the most used open rerankers. It can be loaded with FlagEmbedding's FlagReranker class or with sentence-transformers.
BGE Reranker v2 M3 AI features and use cases
- Reorders retrieved passagesThe model reads the query and each candidate passage together and returns a score. The top passages then go to the language model, which improves answer quality in document search and chat.
- Works across languagesBecause it builds on the multilingual bge-m3, the model can score non-English content. Teams with German, Chinese or mixed documents can use one reranker for all of them.
- Runs on your own infrastructureThe weights are downloadable and the model is described as lightweight with fast inference. Documents never have to leave your servers, which helps with confidential data.
What sets BGE Reranker v2 M3 apart
- Permissive Apache 2.0 license allows commercial use without a separate agreement.
- Very high download numbers on Hugging Face mean many integrations and examples exist.
- No hosted API from the publisher; you run it yourself or through a third-party inference service.
Where BGE Reranker v2 M3 is strongest
China Available internationally
Developed by the Beijing Academy of Artificial Intelligence in Beijing, which also maintains the FlagEmbedding library.
Questions to ask before you buy BGE Reranker v2 M3
You need your own GPU or CPU capacity and someone to run and update the model. Test quality on your own documents and languages before rollout. If you use a third-party inference service instead, its data terms and location apply, not those of BAAI.
BGE Reranker v2 M3 reviews
No reviews yet. Be the first to share how your team uses it.
Every review shows its verification level. We read each one before it goes live. How we check reviews
Questions about BGE Reranker v2 M3
Is BGE Reranker v2 M3 free for commercial use?
Yes, it is released under the Apache 2.0 license, which permits commercial use.
How do I run the model?
Install FlagEmbedding and use the FlagReranker class, or load it with sentence-transformers. It runs on your own hardware.
Does it support German?
The model card describes it as multilingual, based on bge-m3. Test it on your own German documents to judge quality.
BGE Reranker v2 M3 update historyno updates yet
No changes recorded for BGE Reranker v2 M3 yet.
Changes the vendor made and confirmed user reports, each checked against the vendor's own pages. We review every tool monthly.
Alternatives to BGE Reranker v2 M3
Tools in the same category (Rerankers) with the most overlapping markets.
From $3,250 per month · Free trial🇨🇦
Reranking API from Toronto-based Cohere that reorders search results by semantic relevance, multilingual and also for JSON.
From $20 per month · Free plan🇺🇸
Open reranker models from Mixedbread under Apache 2.0 for more than 100 languages and code, also available through the vendor's search API.
Usage-based · Free trial🇩🇪
Embedding and reranker API from Berlin-based Jina AI, part of Elastic since October 2025, with multilingual and multimodal models.