Snowflake Arctic Embed 2.0
Multilingual 568M-parameter embedding model built for retrieval.
About
Snowflake's Arctic Embed 2.0 (the l-v2.0 variant) is a 568M-parameter embedding model designed for high-quality multilingual retrieval without giving up English performance. It supports up to 8,192 tokens via RoPE, uses Matryoshka Representation Learning so vectors compress down to 128 bytes with quantization, and is released under Apache 2.0.
Snowflake reports it beats leading open and proprietary models on MTEB Retrieval, MIRACL, and CLEF — treat those as company numbers until you test on your own data.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Company-reported: tops open/proprietary models on MTEB Retrieval, MIRACL, and CLEF averages.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: strong adoption in multilingual RAG; drop-in replacement story helped; Apache 2.0, 8192-token context via RoPE, Matryoshka compression; benchmark leadership claims are company-reported. License: Apache 2.0 per Hugging Face model card..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · strong adoption in multilingual RAG; drop-in replacement story helped
- · Apache 2.0, 8192-token context via RoPE, Matryoshka compression
- · benchmark leadership claims are company-reported