Skip to main content
Cloud sessions can use any model offered by the workspace’s active LLM providers. A run that names no model uses the workspace default. With no default set, the run is refused with HTTP 428 before any sandbox starts. Before you start: install the CLI, sign in (MUTAGENT_API_KEY=<key> mutagent login --json for a coding agent), select the workspace with mutagent workspaces use <workspace-name>, and add at least one LLM provider to it.

​
mutagent helix models

List the models cloud sessions can use, as provider/model. ★ marks the default.
Pass one of these IDs to mutagent helix --model. An ID that is not in the list is refused with HTTP 422. A line under the list names any LLM provider whose key Helix Cloud cannot use. With --json, the output is { workspaceId, models, defaults, default, unmapped }. Each entry in models has an id to pass as --model. Pick IDs from models[].id; never invent one.
  • models is empty: the workspace has no active LLM provider Helix can use. Add one with mutagent providers add, or copy your local logins with mutagent providers mirror.
  • default is null: runs without --model are refused with HTTP 428. Set a default below.
  • unmapped lists LLM providers whose keys Helix Cloud cannot load into a sandbox.

​
mutagent helix models default

Set the workspace default model. List more IDs after it as fallbacks; they are used in order when the LLM provider of an earlier model has been deactivated. The command replaces the whole list.
Success: exit code 0, and --json prints the same shape as mutagent helix models --json after the change, with your first ID in default. Every ID must appear in mutagent helix models; the server refuses any other ID with HTTP 422 (exit 1). mutagent providers mirror sets a default when the workspace has none. The cloud default does not change your local Helix default. Exit code 3 means you are not signed in or no workspace is selected, and 2 means the key expired or is invalid. See CLI errors.