Also known as: open weights · open source model · local model
An open-weight model is one whose trained parameters are publicly downloadable, allowing anyone to run, inspect, fine-tune, and deploy it on their own infrastructure without calling an external API.
Open-weight is not synonymous with open-source. Most releases publish weights under a licence that may restrict commercial use or specific applications, while withholding training data and code — meaning the result is not reproducible even though it is usable.
The practical appeal is control: data never leaves your infrastructure, costs are predictable, there is no vendor deprecation risk, and fine-tuning is unrestricted. The trade-off is the operational burden of serving models efficiently, which is substantial at scale.
Large Language Model
A large language model (LLM) is a neural network trained on vast quantities of text to predict likely continuations of a sequence, producing systems capable of generating coherent language, answering questions, and writing code.
Fine-Tuning
Fine-tuning is the process of further training a pre-trained model on a smaller, task-specific dataset to adapt its behaviour, style, or domain knowledge without training from scratch.
Inference
Inference is the process of running a trained model to produce output for a given input — as distinct from training, which is the process of creating the model in the first place.
Quantization
Quantization reduces the numerical precision of a model's weights — for example from 16-bit floating point to 8-bit or 4-bit integers — shrinking memory requirements and speeding inference at some cost to output quality.
Knowledge Distillation
Knowledge distillation is a training technique in which a smaller student model learns to reproduce the behaviour of a larger teacher model, yielding a compact model that retains much of the teacher's capability.