Snowflake Arctic
480B-parameter open enterprise MoE that activates only 17B per token.
About
Snowflake Arctic is an open dense-MoE hybrid transformer: 480B total parameters across 128 fine-grained experts with top-2 gating, so only about 17B are active per token. Released under Apache 2.0 with training details published, it was aimed squarely at enterprise workloads like SQL generation and coding.
Its weak spot is context — 4K tokens, small even by 2024 standards — and its size makes self-hosting a serious infrastructure commitment. Snowflake claims it beats DBRX, Llama 3 70B, and Mixtral on enterprise tasks, but those are the vendor's own comparisons.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Snowflake-reported: ahead of DBRX, Llama 3 70B, Mixtral-8x7B on enterprise coding/SQL — vendor's own comparisons, not independent.
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: Major enterprise release with wide press coverage (TechCrunch, DBTA, Flexera); Benchmark wins over DBRX/Llama/Mixtral are Snowflake's own comparisons; Manual review: real open MoE milestone; 4K context is a genuine limitation. License: Apache 2.0.
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
- 8.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · Major enterprise release with wide press coverage (TechCrunch, DBTA, Flexera)
- · Benchmark wins over DBRX/Llama/Mixtral are Snowflake's own comparisons
- · Manual review: real open MoE milestone; 4K context is a genuine limitation