other
0.135B parameters
Commercial OK
Reviewed May 2026

SmolLM2 135M Instruct

SmolLM2-135M-Instruct is the smallest instruction-tuned model in Hugging Face's SmolLM2 family, a 135M-parameter Llama-architecture model trained for on-device deployment. It uses an 8K context window and is shipped with ONNX, GGUF, and Transformers.js artifacts for in-browser inference.

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

Our verdict

OP · Fredoline Eruo|VERIFIED MAY 29, 2026
unrated

The right answer when the constraint is 'must run in a service worker.' SmolLM2-135M is the cleanest open small model we know of for browser and microcontroller-class deployments.

Overview

SmolLM2-135M-Instruct is the smallest instruction-tuned model in Hugging Face's SmolLM2 family, a 135M-parameter Llama-architecture model trained for on-device deployment. It uses an 8K context window and is shipped with ONNX, GGUF, and Transformers.js artifacts for in-browser inference.

Strengths

  • Apache-2.0 and fully open: training data, code, and recipe published
  • ONNX/Transformers.js artifacts run directly in a browser tab
  • Sub-100MB quantized footprint is unmatched for in-RAM inference
  • Llama architecture means zero porting work in any runtime

Weaknesses

  • Reasoning is genuinely weak — do not use for math or multi-step tasks
  • Hallucinates frequently on factual questions
  • 8K context is theoretical; quality degrades well before that
  • No tool-calling template out of the box

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_M0.1 GB1 GB

Get the model

HuggingFace

Original weights

huggingface.co/HuggingFaceTB/SmolLM2-135M-Instruct

Source repository — direct quantization required.

Hardware that runs this

Cards with enough VRAM for at least one quantization of SmolLM2 135M 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 135M Instruct?

1GB of VRAM is enough to run SmolLM2 135M Instruct at the Q4_K_M quantization (file size 0.1 GB). Higher-quality quantizations need more.

Can I use SmolLM2 135M Instruct commercially?

Yes — SmolLM2 135M 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 135M Instruct?

SmolLM2 135M Instruct supports a context window of 8,192 tokens (about 8K).

Source: huggingface.co/HuggingFaceTB/SmolLM2-135M-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 135M Instruct runs on your specific hardware before committing money.