Embeddings

Bulk vector generation for search, RAG, clustering, and deduplication. Common Compute hosts open-weight models; no request touches a third-party API.

Supported models

• bge-large-en-v1.5 — best general-purpose English
• bge-m3 — multilingual, hybrid dense+sparse
• e5-mistral-7b-instruct — instruction-tuned
• nomic-embed-text-v1.5 — long context (8k)

job = cc.embeddings.batch(
    model="bge-large-en-v1.5",
    inputs=stream_from_pg("products"),
)