Usar LetronnaLocal models
Serving your accounts to other tools
The Local endpoint: an OpenAI-compatible address on this computer that other tools can use your connected models through.
Esta página aún no está traducida, así que se muestra en inglés. Puedes traducirla en GitHub. GitHub ↗
The Local endpoint lets other programs on this computer, such as an editor plugin or a script, use the models connected in Letronna through one OpenAI-compatible address and one token, instead of each needing its own keys.
Switch it on
Settings > Accounts, under Local endpoint: switch on "Serve these accounts on this computer". It is off by default. Once on, it shows its address and token, each with a button to copy it.
| Address | http://127.0.0.1:8642/v1 |
| Token | sent as Authorization: Bearer <token>; kept in the system's keychain |
Set local_port in settings.json to use another port.
What it answers
GET /v1/modelslists the models of every connected gateway, each once, local ones included.POST /v1/chat/completionssends a chat to the first gateway that lists the model. If that account is rate-limited, the request is handed on as a session's would be.
Nothing else is served. It listens on 127.0.0.1 only, so other computers cannot reach it, and it answers nothing without the token. Requests through it are paid as your own would be: from your credit on the Letronna gateway, or by the provider whose key it is.