GTE Embeddings

Text embedding models support local semantic search and similarity scoring, with English and multilingual variants and Python or JavaScript inference.

GTE is Alibaba's family of text embedding models for developers building semantic search and text matching into applications. The models turn text into vectors that let an application compare queries with documents by meaning.

Semantic matching: Generate embeddings for sentences or documents and compare their similarity to find related text.
Long text inputs: The English models gte-large-en-v1.5 and gte-base-en-v1.5 accept up to 8,192 tokens per input, so they can process longer passages together.
Multilingual retrieval: The instruction-tuned gte-Qwen1.5-7B-instruct model handles multiple languages and accepts inputs up to 32,768 tokens.
Python and JavaScript support: Run gte-large-en-v1.5 locally with Transformers or SentenceTransformers, or use Transformers.js in JavaScript applications.

These are models to embed in your own application, rather than a standalone search interface. Language coverage and input limits depend on the model you choose. The gte-large-en-v1.5 model is open source under Apache 2.0, and its Python integrations require custom model code.

Embedding and Reranker Models Batch processing Hugging Face integration Multilingual