CPU + 8 GB RAM
FP32 / quantized · FlagEmbedding
The official example runs on CPU; throughput still depends on pair count and sequence length.
BAAI
Embeddings model
BGE Reranker V2 M3
A multilingual reranker that improves the order of search results.
Reranking
Capabilities
What you can use it for.
The most useful reasons to choose this model, without making you read through its repository first.
RerankingMultilingualCross-encoder scoringRAG pipelines
01
RAG pipelines
02
Search ranking
03
Multilingual retrieval
For providers
Want to serve this model?
The provider requirements live here, separate from the information you need to choose and use the model.
Starting pointCPU · 8 GB RAM
Conservative estimate
Source and license
Check the original model.
OpenMayhem links back to the source repository so you can inspect the model card, files, license, limitations, and creator guidance directly.
Keep exploring