other
0.568B parameters
Commercial OK
Reviewed May 2026

Snowflake Arctic Embed L v2.0

Arctic Embed L v2.0 is a 568M-parameter multilingual embedder from Snowflake based on XLM-RoBERTa, producing 1024-dim Matryoshka vectors with an 8192-token context. It is the rare commercial-friendly (Apache-2.0) multilingual model that competes with jina-v3 on cross-lingual MTEB-X while remaining fully redistributable.

License: apache-2.0·Context: 8,192 tokens
BLK · VERDICT

Our verdict

OP · Fredoline Eruo|VERIFIED MAY 29, 2026
unrated

The right pick for any commercial multilingual deployment in 2026. Snowflake released this specifically to fill the license gap left by jina-v3, and the quality is genuinely close. If you need 89-language support and Apache-2.0, this is effectively the only choice at this parameter count.

Overview

Arctic Embed L v2.0 is a 568M-parameter multilingual embedder from Snowflake based on XLM-RoBERTa, producing 1024-dim Matryoshka vectors with an 8192-token context. It is the rare commercial-friendly (Apache-2.0) multilingual model that competes with jina-v3 on cross-lingual MTEB-X while remaining fully redistributable.

Strengths

  • Apache-2.0 multilingual at ~568M — currently the best license/quality tradeoff above jina-v3 for commercial use
  • 1024-dim Matryoshka truncatable to 256 dims with <3% nDCG@10 loss on BEIR
  • Strong cross-lingual MTEB-X (~54) — English queries retrieve foreign-language documents reliably
  • Native 8K context via XLM-RoBERTa with extended RoPE

Weaknesses

  • 568M params — heavier than nomic-v1.5 on CPU; needs GPU or ONNX for sub-100ms latency at scale
  • Pure English MTEB (~58) trails English-only specialists like mxbai/bge-large
  • No GGUF in repo at launch — community quantizations only
  • Newer than BGE/jina ecosystem so fewer downstream fine-tunes available

Quantization variants

Each quantization trades model quality for file size and VRAM. Q4_K_M is the most popular starting point.

QuantizationFile sizeVRAM required
Q4_K_M0.3 GB1 GB

Get the model

HuggingFace

Original weights

huggingface.co/Snowflake/snowflake-arctic-embed-l-v2.0

Source repository — direct quantization required.

Hardware that runs this

Cards with enough VRAM for at least one quantization of Snowflake Arctic Embed L v2.0.

Compare alternatives

Models worth comparing

Same parameter band, plus what's one tier above and below — so you can decide what actually fits your hardware.

Step down
Smaller — faster, runs on weaker hardware
No verdicted models in the next tier down yet.

Frequently asked

What's the minimum VRAM to run Snowflake Arctic Embed L v2.0?

1GB of VRAM is enough to run Snowflake Arctic Embed L v2.0 at the Q4_K_M quantization (file size 0.3 GB). Higher-quality quantizations need more.

Can I use Snowflake Arctic Embed L v2.0 commercially?

Yes — Snowflake Arctic Embed L v2.0 ships under the apache-2.0, which permits commercial use. Always read the license text before deployment.

What's the context length of Snowflake Arctic Embed L v2.0?

Snowflake Arctic Embed L v2.0 supports a context window of 8,192 tokens (about 8K).

Source: huggingface.co/Snowflake/snowflake-arctic-embed-l-v2.0

Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify model claims.

Related — keep moving

Before you buy

Verify Snowflake Arctic Embed L v2.0 runs on your specific hardware before committing money.