Openshell Inference set multiple models

[bmahabirbu]

Problem Statement

a provider like claude or a local one can have multiple models associated with a single endpoint. The inference argument should have the ability to set --model model1 model2

then in the sandbox implicitly it will switch a model based on the model name

Proposed Design

openshell inference set --model1 model2

Alternatives Considered

a method to switch models implicitly intead of having to reset the inference update

Agent Investigation

No response

Checklist

• I've reviewed existing issues and the architecture docs
• This is a design proposal, not a "please build this" request

[mogul]

See also #2039

[johntmyers]

📋 triage-agent

Closing as superseded by #3195. The openshell inference set command and single managed route model were removed. Sandboxes now call attached provider native endpoints, where each request can use the provider model semantics directly rather than configuring a model list in OpenShell.