44 lines
1.4 KiB
Markdown
44 lines
1.4 KiB
Markdown
---
|
|
title: Chats
|
|
description: Jan is a ChatGPT-alternative that runs on your own computer, with a local API server.
|
|
keywords:
|
|
[
|
|
Jan,
|
|
Rethink the Computer,
|
|
local AI,
|
|
privacy focus,
|
|
free and open source,
|
|
private and offline,
|
|
conversational AI,
|
|
no-subscription fee,
|
|
large language models,
|
|
]
|
|
---
|
|
|
|
:::caution
|
|
|
|
This is currently under development.
|
|
|
|
:::
|
|
|
|
## Overview
|
|
|
|
In Jan, `chats` are LLM responses in the form of OpenAI compatible `chat completion objects`.
|
|
|
|
- Models take a list of messages and return a model-generated response as output.
|
|
- An [OpenAI Chat API](https://platform.openai.com/docs/api-reference/chat) compatible endpoint at `localhost:1337/v1/chats`.
|
|
|
|
## Folder Structure
|
|
|
|
Chats are stateless, thus are not saved in `janroot`. Any content and relevant metadata from calling this endpoint is extracted and persisted through [Messages](/docs/engineering/messages).
|
|
|
|
## API Reference
|
|
|
|
Jan's Chat API is compatible with [OpenAI's Chat API](https://platform.openai.com/docs/api-reference/chat).
|
|
|
|
See [Jan Chat API](https://jan.ai/api-reference/#tag/Chat-Completion)
|
|
|
|
## Implementation
|
|
|
|
Under the hood, the `/chat` endpoint simply reroutes an existing endpoint from [Nitro server](https://nitro.jan.ai). Nitro is a lightweight & local inference server, written in C++ and embedded into the Jan app. See [Nitro documentation](https://nitro.jan.ai/docs).
|