Architecture
Billion Parameters
The standard unit for expressing model scale, where larger usually means more capable.
Modern LLMs are described by their parameter count in billions (B). A 7B model is fast and cheap to run locally; a 405B model rivals frontier closed APIs but requires substantial GPU clusters. Parameter count is a rough proxy for capability — architecture, training data quality, and fine-tuning often matter more than raw size.
Related Terms
Parameter
A learnable weight in a neural network; model size is measured in billions of parameters.
LLM
Large Language Model — a neural network trained on vast text corpora to generate human-like text.
Quantization
Compressing model weights to lower numerical precision to reduce memory and speed up inference.
Mixture of Experts (MoE)
An architecture where only a subset of model parameters is activated per token.