What would you like to be added?
Allow for bypassing the interactive model picker by specifying the model name/id in the command, as in /model qwen3-coder-next.
Why is this needed?
When connecting to a local proxy which provides several models, it is simplest to configure qwen code with the proxy's base url and launch with --model. This avoids the need to update the local configuration every time the upstream server adds/deletes/updates a model. Configuring it this way, however, also makes it impossible to switch between models mid-session, because the interactive selector only shows models which are explicitly configured in the local config. Allowing the user to specify the model name would bypass this issue.
Additional context
Claude already supports this as a default behavior alongside its interactive model selector.
I am willing to take on implementing this if the feature is wanted.
What would you like to be added?
Allow for bypassing the interactive model picker by specifying the model name/id in the command, as in
/model qwen3-coder-next.Why is this needed?
When connecting to a local proxy which provides several models, it is simplest to configure qwen code with the proxy's base url and launch with
--model. This avoids the need to update the local configuration every time the upstream server adds/deletes/updates a model. Configuring it this way, however, also makes it impossible to switch between models mid-session, because the interactive selector only shows models which are explicitly configured in the local config. Allowing the user to specify the model name would bypass this issue.Additional context
Claude already supports this as a default behavior alongside its interactive model selector.
I am willing to take on implementing this if the feature is wanted.