Retrieval models we evaluate for Arabic documents: multilingual and Arabic-specific embeddings, plus rerankers. Curated, not endorsed.
-
BAAI/bge-m3
Sentence Similarity • Updated • 33.7M • • 4.03k -
Qwen/Qwen3-Embedding-8B
Feature Extraction • 8B • Updated • 2.12M • • 862 -
Qwen/Qwen3-Embedding-0.6B
Feature Extraction • 0.6B • Updated • 7.41M • • 1.36k -
intfloat/multilingual-e5-large-instruct
Feature Extraction • 0.6B • Updated • 1.83M • • 660