Quickstart
Your first ten minutes with Rin Chat
Rin Chat runs on your computer and talks to an AI provider you choose. Getting from download to first reply takes the five steps below, and only one of them, connecting a provider, needs any thought.
Not sure which AI to connect? Read Choosing a provider
Install
Get the installer for Windows or Linux from the download page and run it. After that, Rin Chat tells you when a new version is out and installs it when you say so.
Make your local account
The first screen asks for a username and a password. This account lives only on your computer: the password is what unlocks your encrypted library, and nothing is sent to us.
Save your recovery kit now
There is no password reset. If you forget the password, the only ways back in are your recovery kit (24 words, or the file holding them) or a linked aicharactercards.com account. Open Settings → Data & Backup → Recovery kit and keep a copy somewhere safe, outside the computer it protects.
Want to skip the password on every launch? Turn on Remember me on this device, or Windows Hello, in Settings → Account & Security.
Connect an AI provider
Rin Chat does not ship a model. A provider is the AI that writes the replies: a paid service, a free one, or a model running on your own computer.
- Open Settings → LLM Providers and click + Add provider.
- Pick your provider from the Type list and paste its API key. The address is filled in for you.
- Click Test. It tells a wrong address from a wrong key in one step.
- Save. Rin Chat makes a generation config for it and makes it active. Open it in Settings → Generation Configs to pick the model.
No key yet? Z.AI has three models that are free to use, and LM Studio or Ollama run a model on your own computer with no key at all. Choosing a provider compares the options.
Start a chat
Open the Library. Two characters are already there: Jinx Wilder, and the Adventure Mode Narrator, a game master with real dice. Open one, click New chat, type a message and press Enter.
Any reply can be edited or regenerated, and regenerating keeps the old versions to step back through with the arrow keys, and the model name in the chat's top bar switches models mid-chat.
Add your own characters
- Import a card. Drag a
.pngor.jsoncard onto the Library, or use Import. SillyTavern cards work as they are. - Import from a link. Paste a card's page from Chub, RisuRealm, Pygmalion, Perchance or aicharactercards.com.
- Build one with Card Forge. Under Tools, describe the character you want and an AI guide builds the card with you. See Card Forge.
- Write one by hand. + New in the Library opens the card editor. The card format guide explains every field.
Finding your way around
- The title bar holds everything: Library for characters, lorebooks and personas; Tools for Card Forge, Image Studio, Story Maker and Direct Chat; Guide for the manual.
Ctrl+Ksearches the whole app from anywhere, settings included.- Settings opens in Basic. Turn on Show advanced settings at the top of the Settings list to see everything.
- The Guide opens with Take the app tour, a short look around whenever you want it.
When something goes wrong
| You see | It means |
|---|---|
| An empty model list, or "Failed to fetch" | The address is wrong or the server is not running. Local servers also need CORS turned on (LM Studio: Developer tab, server settings). |
| 401 or 403 | The API key is wrong or missing. |
| 402 | The account has no credit. Services like OpenRouter and NanoGPT need credit added before the first message. |
| 404 | The address is missing its path. Most end in /v1. |
| "Temporarily overloaded" | The provider is busy. Wait, or switch models from the button under the error. |
The troubleshooting section of the guide covers the rest, and the Discord is the place to ask.
Not sure which AI to connect? Read Choosing a provider