Live answers · updated

What are the most used open embedding models?

The most downloaded open embedding model on Hugging Face in the last 30 days is BAAI/bge-small-en-v1.5, with 49,421,693 downloads (license: mit).

BAAI/bge-small-en-v1.549,421,693 downloads · 662 likes · mit
BAAI/bge-large-en-v1.59,103,164 downloads · 781 likes · mit
Qwen/Qwen3-Embedding-0.6B8,958,817 downloads · 1,323 likes · apache-2.0
BAAI/bge-base-en-v1.58,140,331 downloads · 540 likes · mit
intfloat/multilingual-e5-large6,813,094 downloads · 1,282 likes · mit
ibm-granite/granite-embedding-small-english-r26,753,887 downloads · 113 likes · apache-2.0
Xenova/all-MiniLM-L6-v24,551,539 downloads · 179 likes · apache-2.0
BAAI/bge-small-zh-v1.53,238,510 downloads · 166 likes · mit
BAAI/bge-base-en2,613,126 downloads · 73 likes · mit
Xenova/bge-base-en-v1.52,452,705 downloads · 28 likes · mit

Source: Hugging Face Hub public API (downloads in the last 30 days). Recomputed every few minutes.

Get this by API

One call, US$0.002 in USDC (x402, Base or Solana), or with an X-API-KEY. 100 free calls/day with the header X-Free-Trial: 1.

curl -H "X-Free-Trial: 1" "https://intel.degenscan.io/v1/ai/models/by-task?task=embeddings&limit=10"

MCP: https://intel.degenscan.io/mcp → tool ai_models_by_task. Docs: https://intel.degenscan.io/docs · llms.txt

Related

Market data and analytics only — not investment advice. Degenscan Intel · Marbella Collins LLC · contact@degenscan.io