Snowflake Arctic Instruct
480B dense-MoE hybrid with 17B active parameters, Apache 2.0.
About
Snowflake Arctic Instruct combines a 10B dense transformer with a residual 128x3.66B MoE MLP, yielding 480 billion total parameters with only 17 billion active via top-2 gating. It was pre-trained from scratch by the Snowflake AI Research Team.
Released under Apache 2.0 with both base and instruct checkpoints, Arctic is positioned for enterprise AI, SQL generation, and coding tasks. The team published cookbooks for training custom MoE models alongside the release.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
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: Verified via vendor site/docs/GitHub and reputable coverage, wave 134; Trending score 80/100 from researcher assessment.