Snowflake Open Sources Arctic, Family of Embedding Models for RAG

Snowflake and Databricks are increasingly becoming Generative AI Native

Hello Everyone,

I’ve been fairly impressed with Databricks and their models recently and Snowflake is also making similar investments. Working with Mistral they developed a Snowflake Copilot that looks somewhat useful.

This week it’s come out that they are also open sourcing their best embedding models for the world! They are now available open source in HuggingFace with an Apache 2 license.

These models are better than you may have realized. The largest model in the family, with 330 million parameters, leads the Massive Text Embedding Benchmark (MTEB) Retrieval Leaderboard, achieving an average retrieval performance surpassing 55.9.