bge-base-en-v1.5

The Xenova/bge-base-en-v1.5 model is a general-purpose English text embedding model, converted to ONNX for efficient deployment. It is based on BAAI’s BGE-base-en-v1.5, a widely used retrieval-oriented embedding model. The model outputs 768-dimensional dense vectors and is designed for semantic similarity, retrieval, and classification tasks.

Key capabilities

• Computes sentence-level embeddings with mean pooling and L2 normalization.
• Supports retrieval with a recommended query instruction prefix: Represent this sentence for searching relevant passages: .
• ONNX format ensures low-latency inference suitable for production APIs.

Best for

• Semantic search and document retrieval.
• Clustering, classification, and deduplication of English text.
• Applications requiring a lightweight base-sized embedding model (768 dimensions).

Performance

As a base-sized BGE model, it offers a strong balance of speed and accuracy for English embedding tasks. The original BAAI/bge-base-en-v1.5 has been benchmarked on MTEB and other retrieval leaderboards (exact numbers are not included in this card; refer to the BAAI model page for detailed scores).

Usage with gigarouter

Gigarouter hosts this model as a managed, OpenAI-compatible API. No local installation or ONNX conversion is required. The API accepts text input and returns normalized embeddings suitable for cosine similarity or other vector operations.