Use a model in BCCLI

A model in BCCLI is always a provider/model pair. The provider decides where the work happens — BotConnector Cloud, or a runtime on this machine — and the model decides which one answers.

See what is reachable

bccli models

The list is built from the providers this machine is signed in to, so it is short and true rather than the whole catalogue. If a model you expect is missing, the provider is not configured here yet.

Pin a model

bccli -m provider/model

bccli -m bc-cloud/glm-5.3-flash

Without -m, the session keeps whatever provider and model it used last. Inside a session you can switch instead of restarting: /model and /provider.

One task, one model

-m combines with -p, which is what a script or a CI job wants — both the task and the model stated up front, nothing inherited from a previous session:

bccli -m bc-cloud/glm-5.3-flash -p "summarise the open TODOs in src/"

Choosing between cloud and local

Use a cloud model when you want the catalogue's hosted models and per-token billing. Use a local one when the code must not leave the machine, or when you already run an open model yourself. Both appear in the same bccli models list; they are separate providers behind it.

Next

• BCCLI models — managing the providers behind the list.
• Model catalog — context length and price per hosted model.