other
0.36B parameters
Commercial OK
Reviewed May 2026

SmolLM2 360M Instruct

SmolLM2-360M-Instruct is the middle tier of the SmolLM2 instruct family, a 360M-parameter Llama-architecture model with an 8K context. It is shipped with ONNX and Transformers.js artifacts and aimed at on-device assistants that need more capability than the 135M can deliver.

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

Our verdict

OP · Fredoline Eruo|VERIFIED MAY 29, 2026
unrated

A sensible step up from 135M when you have a bit more silicon. The most defensible choice when you need to fine-tune a tiny model on private data and ship it on hardware without a GPU.

Overview

SmolLM2-360M-Instruct is the middle tier of the SmolLM2 instruct family, a 360M-parameter Llama-architecture model with an 8K context. It is shipped with ONNX and Transformers.js artifacts and aimed at on-device assistants that need more capability than the 135M can deliver.

Strengths

  • Roughly 2-3x more useful than the 135M for the same deployment class
  • Apache-2.0, fully open training pipeline
  • Multiple quantizations (q4f16, q8, bnb4) prebuilt in the repo
  • Tiny enough to fit on a Raspberry Pi 5 with headroom

Weaknesses

  • Still trails Qwen3-0.6B on most benchmarks despite similar size
  • 8K context, no GQA optimizations beyond stock Llama
  • Limited community fine-tunes compared to Qwen/Llama tiers
  • No native tool-use template

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_M0.2 GB1 GB

Get the model

HuggingFace

Original weights

huggingface.co/HuggingFaceTB/SmolLM2-360M-Instruct

Source repository — direct quantization required.

Hardware that runs this

Cards with enough VRAM for at least one quantization of SmolLM2 360M Instruct.

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 up
More capable — bigger memory footprint
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 SmolLM2 360M Instruct?

1GB of VRAM is enough to run SmolLM2 360M Instruct at the Q4_K_M quantization (file size 0.2 GB). Higher-quality quantizations need more.

Can I use SmolLM2 360M Instruct commercially?

Yes — SmolLM2 360M Instruct ships under the apache-2.0, which permits commercial use. Always read the license text before deployment.

What's the context length of SmolLM2 360M Instruct?

SmolLM2 360M Instruct supports a context window of 8,192 tokens (about 8K).

Source: huggingface.co/HuggingFaceTB/SmolLM2-360M-Instruct

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

Related — keep moving

Before you buy

Verify SmolLM2 360M Instruct runs on your specific hardware before committing money.