# Multi-Turn Conversation

> A chat with many messages back and forth. The model rereads the whole chat each turn.

**Field:** Artificial Intelligence  
**Category:** Working with AI  
**Also known as:** Multi-Turn Chat, Conversation History (AI), Chat History

## Think of it like

It is like a game of catch. Each throw builds on the last one.

## Definition

A multi-turn conversation is a chat with many back-and-forth messages. Each time you ask and the model replies, that is one turn. The model has no memory of its own between turns. So the app sends the whole chat so far with every new message. That is how the model knows what "it" or "that one" means. Long chats fill the context window and cost more.

## Example

Emma asks for a cake recipe. Then she types "make it for 12 people". The model knows which recipe she means, because the first message is sent again.

## Example code

```json
[
  { "role": "user", "content": "Give me a cake recipe." },
  { "role": "assistant", "content": "Here is a lemon cake..." },
  { "role": "user", "content": "Make it for 12 people." }
]
```

Turn two only makes sense because turn one is sent along with it.

## Why it matters

It is why a chat feels like it remembers. It is also why a very long chat gets slow, dear and muddled.

## Common mistake

Thinking the model learns from the chat. It only rereads it, and it forgets it all once the chat ends.

## Three principles

- **Turns add up.** Each new message joins the pile.
- **Sent every time.** The whole chat goes with each turn.
- **Start fresh.** A new chat clears a muddled one.

## Related terms

- [Prompt (AI)](https://learn.benibix.com/term/ai/prompt)
- [Message Roles](https://learn.benibix.com/term/ai/message-roles)
- [Context Window](https://learn.benibix.com/term/ai/context-window)
- [Token (AI)](https://learn.benibix.com/term/ai/token)
- [Chatbot](https://learn.benibix.com/term/ai/chatbot)
- [Context Engineering](https://learn.benibix.com/term/ai/context-engineering)

## Taught in

- [Multi-Turn Conversations](https://learn.benibix.com/learn/ai-working-with-ai/multi-turn-conversations): Prompt Engineering, 2/16

![A pile of messages sent to the model on each turn. At the top, custom instructions. Under it, the first question and its reply, then the new question. An arrow takes the whole pile to the model, which writes the new reply. A regenerate arrow loops back from the newest reply to draw another one.](https://learn.benibix.com/diagrams/ai3-chat-turns.svg)

---

Source: https://learn.benibix.com/term/ai/multi-turn-conversation
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
