Local LLM
Model Lokal
Running the model on your own machine or server, with no provider in the loop.
A local model runs on hardware you control: a laptop, a workstation, or a server in your own data centre. Nothing about your prompts or documents leaves that machine.
The reasons to go local are real: data that cannot leave the organisation, models that need to be fine-tuned on private data, predictable cost at high volume, and no dependency on a provider's availability. The reasons not to are equally real: the hardware cost, the operational work of keeping it running, and the fact that the best model for a hard task is usually the one you cannot afford to host.
The middle path is common and underrated: run a local model for the high-volume, low-stakes work, and send the hard cases to a hosted model. Splitting by difficulty is usually cheaper than picking one side.
Related terms
Quantization
Storing weights in fewer bits so a model runs with less memory and cost.
Model
The trained file of learned parameters that turns input into output.
Cost per Token
What you pay for each unit of text the model reads or writes.
Privacy
A property of information: who can see it, who can use it, and who cannot be told.
Model Provider
The company that serves a model to you. Usually someone else's server.
Want this working inside your team?
It starts with one conversation about your workflow.