# What is Autoregressive Model? Definition & Technical Mechanics

> **Source:** [SPIDITS AI AI Glossary](https://spidits.com/ai-glossary)  
> **Category:** Foundational AI  
> **Canonical Citation:** [https://spidits.com/ai-glossary/autoregressive-model](https://spidits.com/ai-glossary/autoregressive-model)

## Definition
An Autoregressive Model is an AI model that predicts future values in a sequence based on past values. In LLMs, autoregressive generation works by taking the prompt, predicting the next word, appending that word to the prompt, and repeating the process.

## Architectural Mechanics & Deep Dive
An autoregressive model is a sequence generation architecture that predicts the next token in a sequence based on all previously generated tokens. In Large Language Models, autoregressive decoding generates text one word at a time, appending the newly predicted word to the context window before calculating the probability distribution for the next. This step-by-step, sequential nature makes generation highly coherent but computationally expensive.

## Industry Applications & Real-World Use Cases
Text generation, weather forecasting, and stock price prediction

## Frequently Asked Questions

### How does autoregressive decoding work?
The model outputs one token at a time. Each new token depends on all the tokens that came before it.

### What are the alternatives to autoregressive generation?
Non-autoregressive generation outputs the entire sequence at once, which is faster but often has lower quality.

## Related Terminology
- [llm](https://spidits.com/ai-glossary/llm)
- [transformer](https://spidits.com/ai-glossary/transformer)
- [tokenization](https://spidits.com/ai-glossary/tokenization)

---
*From the SPIDITS AI Knowledge Base (336+ terms indexed). Explore full technical definitions: [https://spidits.com/ai-glossary](https://spidits.com/ai-glossary)*