Conversation History
Also called Chat History
The running transcript of a chat that gets replayed to the model so it stays coherent.
Think of it like
Re-reading the whole email thread before you reply so you do not repeat yourself.
Example
Each time you send a message, the app resends the prior turns so the assistant "remembers" your name from earlier.
How it actually works
Because models are stateless, the app rebuilds the conversation each turn by prepending previous messages. This works until the history outgrows the context window, at which point you must trim old turns, summarize them, or retrieve only the relevant bits. Naively keeping everything is the most common cause of ballooning token costs.
For product teams
Every remembered detail is a resent token you pay for; long chats quietly get expensive and slow.
For engineers
Stateless inference re-hydrated per request; token budget forces truncation, summarization, or selective retrieval.
Related
- Short-Term Memory — Fills the model’s short-term memory.
- Summarization Memory — Gets compressed once it grows too long.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome