Jina Embeddings v2 Base EN

Jina Embeddings v2 Base EN (jinaai/jina-embeddings-v2-base-en) is a 137-million-parameter English text embedding model released in October 2023. Based on the JinaBERT architecture with symmetric bidirectional ALiBi, it was the world's first open-source embedding model supporting an 8,192-token sequence length, trained on 400M+ sentence pairs and hard negatives.

It produces 768-dimensional vectors and performs comparably to OpenAI's text-embedding-ada-002 on standard benchmarks at a fraction of the size (0.27GB). Released under Apache 2.0, it remains a popular lightweight choice for long-document retrieval, RAG, and semantic search.