# Latency

> How long an AI system takes to begin or finish a response, tool call, or workflow.

- Category: Reasoning & Inference
- Stage: Established
- Also called: Response time, Time to first token
- Canonical: https://www.joelcsaji.com/ai-dictionary/latency

## Plain English

It is the wait time users feel.

## Example

A reasoning model may produce a better answer but take longer before the first streamed token appears.

## Why it matters

AI product design constantly trades latency against quality, cost, safety checks, and reasoning depth.

## Related

- [Streaming](https://www.joelcsaji.com/ai-dictionary/streaming)
- [Reasoning Effort](https://www.joelcsaji.com/ai-dictionary/reasoning-effort)
- [Model Routing](https://www.joelcsaji.com/ai-dictionary/model-routing)

## Mentioned in

- [Fable 5, Export Controls, and GLM 5.2](https://www.joelcsaji.com/blog/fable-5-export-controls-and-glm-5-2)

## Sources

- [OpenAI: Streaming Responses](https://platform.openai.com/docs/guides/streaming-responses)
- [OpenAI: Background Mode](https://developers.openai.com/api/docs/guides/background)