# Quantization

> Reducing numerical precision in a model to make it smaller, faster, or cheaper to run.

- Category: Models & Training
- Stage: Established
- Also called: Low-bit inference, Model compression
- Canonical: https://www.joelcsaji.com/ai-dictionary/quantization

## Plain English

The model uses lighter-weight numbers.

## Example

A developer runs a quantized open model locally on a laptop GPU.

## Why it matters

Quantization makes local and edge AI more practical, especially for smaller devices.

## Related

- [Small Language Model](https://www.joelcsaji.com/ai-dictionary/small-language-model)
- [Latency](https://www.joelcsaji.com/ai-dictionary/latency)
- [Distillation](https://www.joelcsaji.com/ai-dictionary/distillation)

## Sources

- [Stanford HAI: 2026 AI Index](https://hai.stanford.edu/ai-index/2026-ai-index-report)