How do you choose an embedding model?

medium

Answer

$(1) MTEB leaderboard benchmark.$
$(2) Task fit: retrieval / clustering / classification differ.$
$(3) Dimension: 768 / 1024 / 1536 / 3072 — bigger = better + more storage.$
$(4) Language: multilingual (mE5, BGE-M3) if needed.$
$(5) Cost: OpenAI ada-3 SaaS vs OSS (BGE, Voyage).$
$(6) Fine-tune on domain data for +10-20% gains.$
$(7) License.$
$Modern top: OpenAI text-embedding-3-large, Voyage-3, BGE-M3, Nomic-embed, Cohere-v3.$
$Evaluate on your own retrieval benchmark.$

Check yourself — multiple choice

Just one
MTEB benchmark + task fit + dimension (768-3072) + language multilingual + cost SaaS vs OSS + fine-tune domain (+10-20%) + license; top: text-embedding-3-large / Voyage-3 / BGE-M3 / Nomic; eval on own bench
Random
Not real

Embedding choice: MTEB + task + dim + lang + cost + fine-tune + own bench.

#llmops

215 interview questions in this topic.