Configure models and endpoints

Connect a model runtime and understand where requests go.

Applies to Bridge rc.32

Choose the endpoint

Open the model/runtime controls in Settings → Configuration. Add or select the endpoint you intend to use, such as Ollama on the computer or vLLM on your network. Local and LAN endpoints are the default path.

Confirm the model

Refresh the endpoint’s model list and select an available model. A saved model name is not proof that its server is running or that it fits in memory. Check a small non-sensitive request after setup. Vision also needs a model that supports the attached images.

Public endpoints change the boundary

An explicitly confirmed public endpoint may receive prompts, tool descriptions and requests outside your network. The Alluqi account is not an inference proxy. Only add a provider credential through the desktop’s own setup controls; never paste it into chat or support.

Context and failures

The context indicator is an estimate. Compaction can shorten the working context while saved history remains available. On a connection or capacity failure, check the configured endpoint and current model rather than assuming Bridge changed models for you.