SmolLM2-360M
Hugging Face's 360M tiny model that punches above its weight on-device.
About
SmolLM2-360M is the smallest member of Hugging Face's SmolLM2 family, released in late October 2024 under Apache 2.0. It was trained with a data-centric multi-stage recipe on roughly 11 trillion tokens spanning web text, math, code, and instruction data, giving it unusually strong reasoning for a 360M parameter model.
The model targets smartphones and edge devices, with an 8K context window and BF16 weights under 1GB. An official instruct variant fine-tuned with SFT and DPO is also available, and the family is a popular default for local-first apps and air-gapped deployments.
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 124; Trending score 78/100 from researcher assessment.