Home / Glossary / Parameters
What is Parameters?
Parameters are the learned weights of a neural network. osFoundry’s catalog labels every model with its parameter count (e.g. 7B, 70B) so you can match model size to your hardware and quality needs.
Detail
Parameter count is the rough proxy for an LLM’s capacity. A 7B model has 7 billion learned weights; a 70B has 70 billion. More parameters generally means better quality but exponentially more memory and compute. A 7B Q4 model needs ~6 GB VRAM; a 70B Q4 needs ~50 GB.
Parameter count is not the only quality factor — training data, architecture, and post-training matter too. Modern 7B models often beat older 70B models on common benchmarks.
Related terms
Related features