Context

At the moment, the editor and weak models can be set in the config or in a the command line arguments.

But, once aider is started:

• The editor model cannot be changed explicitly (but it does change when the main model is changed via /model command)
• The weak model cannot be changed at all

(Please, correct me if this is possible and I simply missed how to do it)

This is especially inconvenient in the current state of things, with DeepSeek-R1 being very good in combination with sonnet editor, but DeepSeek having random outages.

Here is one particular situation that happens often:

• Aider is started with R1 architect and Sonnet editor, which implicitly sets deepseek-chat as the weak model
• Aider is being used and at some point DeepSeek becomes unresponsive
• Switching to sonnet with /model changes the main and editor models, but the weak still uses deepseek, which is not working, so aider must be restarted with new arguments
• If aider is restarted and at some point DeepSeek comes back online, switching back to R1 + Sonnet from inside aider is not possible either, because switching the main model to R1 also switches the editor to deepseek-chat

Feature Request

Add 2 new in-chat commands /editor-model and /weak-model that would give the user full control about which models to use. The /model command can either continue to work as it does now, switching the editor model to the most appropriate companion (and maybe then the weak model should also be changed?), or just not touch the editor model at all and let the user do that explicitly.