What is a Small language model?
A small language model (SLM) is a language model with far fewer parameters than frontier LLMs, small enough to run on modest hardware or inside your own network.
In more detail
There is no formal cutoff, but SLMs usually run from under one billion to around ten billion parameters, against hundreds of billions for the largest hosted models. They are less capable in general. After fine-tuning on a narrow task, though, one can match a larger model at a fraction of the cost and delay. Because they can run on your own servers, they suit data you cannot send to an outside API. The trade-off is upkeep: you own the hosting, the updates and the evaluation. For open-ended reasoning or messy text, a large model still wins. For high-volume, well-defined work such as sorting documents, an SLM is often the better engineering choice.
Example
A law firm's intake team routes a few thousand inbound documents a week by matter type. A fine-tuned SLM on the firm's own server does the routing, and no client text leaves the network.