Learn AI · Level 3 · Tech and tools · step 2 of 16 · word 3 of 7
What is a small language model (SLM)?
A small language model is a compact AI model, often a few billion parameters, that can run cheaply on a laptop or phone. It is less capable than the biggest models, but faster and more private.
In plain English
The biggest chatbots run on huge models in data centres. Small language models are slimmed-down versions, with a few billion parameters instead of hundreds of billions. That makes them cheap to run and small enough for ordinary devices.
Examples include Microsoft’s Phi models and Google’s Gemma models. Apple’s on-device model in iPhones has about 3 billion parameters.
An example
Your phone summarises a long notification or suggests a reply without sending anything to the internet. A small model running on the phone itself does that.
Watch out
Small models know less and make more mistakes on hard questions. They are best for focused jobs such as sorting, summarising or answering from your own documents. Many are trained by distillation from bigger models.
Learn AI path · Level 3 · Tech and tools · step 2 of 16
Next wordPrompt engineering →See all of level 3