Your first inference session
Choose a model, bring its contributors online, and start a conversation. You can use the same session from Ygaraty’s chat page or from your own application.
Start here¶
Sign in with your Google or GitHub account.
Create a session and invite the contributors it needs.
Wait until Open model chat becomes available.
An inference session runs an existing model to answer your messages. Its distribution plan specifies how many contributors are needed and the GPU memory each one supplies. A contributor is a browser running part of that model.
You need the Ygaraty panel address provided by your group or service operator. The available models, plans and connection options depend on that service. You do not need to install Ygaraty to use the panel or chat.
Keep these three things separate¶
| Item | What it is for |
|---|---|
| Your account | Creating and managing your sessions in the panel |
| Contributor invitation link | Letting someone join your session and supply compute |
| Session ID | Letting an application use that session for inference |
Share the invitation link and session ID only with people you intend to give access. Anyone with the session ID can use its inference API; no separate API secret is required.
If a step is unavailable, see Help.