BGE Large
Available nowRetrieve focused English passages when search relevance matters more than a smaller index.
What it does
BGE Large English v1.5 is the larger English member of BAAI’s BGE v1.5 family. It maps questions and passages into vectors so a search system can retrieve information with similar meaning.
Compared with BGE Base, it uses a larger model and 1,024-dimensional vectors. It is suited to evaluating more demanding English retrieval tasks while keeping the same short-passage approach. It has a 512-token source context, rather than the longer context of BGE-M3.
Where it helps
Policy lookup
Find the paragraph that answers a specific English policy question.
Support assistants
Retrieve relevant FAQ passages before an assistant prepares an answer.
Editorial archives
Find related descriptions and short sections of English articles.
When to choose it
Try it when your collection is English, your passages are short, and Base misses relevant matches. Judge the improvement on your own documents and queries rather than assuming a larger model will always win.
It costs more here than BGE Base or BGE-M3. It does not replace document chunking or add multilingual coverage; the quality of the indexed passages still matters.
Official sources
Paraphrased from the provider’s documentation and original model information. Selection advice is our editorial guidance.
BAAI · BGE Large English v1.5 model card ↗Reviewed October 3, 2026