# What is Speculative Decoding? Definition & Technical Mechanics

> **Source:** [SPIDITS AI AI Glossary](https://spidits.com/ai-glossary)  
> **Category:** Model Training  
> **Canonical Citation:** [https://spidits.com/ai-glossary/speculative-decoding](https://spidits.com/ai-glossary/speculative-decoding)

## Definition
Speculative Decoding is a latency optimization technique that accelerates LLM generation. A smaller, faster drafting model proposes multiple candidate tokens, which are then validated in parallel by the larger target model in a single forward pass.

## Architectural Mechanics & Deep Dive
Speculative decoding is an inference acceleration technique that uses a small, fast draft model to speculate multiple tokens ahead, and then verifies them in parallel using a large target model in a single forward pass. This mathematical shortcut achieves identical target outputs while accelerating token generation speeds.

## Industry Applications & Real-World Use Cases
API throughput optimization, interactive chatbot responses, and inference cost reduction.

## Frequently Asked Questions

### Does speculative decoding alter the quality of the output?
No, speculative decoding mathematically guarantees the exact same token probability distribution as the target model alone.

### What speedup is expected from speculative decoding?
It typically increases token generation speed by 2x to 3x depending on the alignment between the draft and target models.

## Related Terminology
- [inference](https://spidits.com/ai-glossary/inference)
- [distillation](https://spidits.com/ai-glossary/distillation)
- [token](https://spidits.com/ai-glossary/token)

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