Allow users to set a default model and reasoning level for new Work chats

What variant of Codex are you using?

ChatGPT Desktop App for macOS — Work mode

What feature would you like to see?

Please add a user-level setting that allows us to choose the default model and reasoning level used when starting a new Work chat.

Currently, it is very easy to start a new Work chat and forget to change the model before submitting the first prompt. If Work defaults to a high-cost model or reasoning level, even a relatively trivial task can consume a significant portion of the shared Work/Codex usage allowance.

Ideally, Settings would provide defaults such as:

• Default Work model
• Default Work reasoning level
• Optionally, "remember last used model/reasoning level"

These defaults should apply automatically whenever a new Work chat is created, while still allowing the model and reasoning level to be changed for an individual chat.

Additional information

The motivation is avoiding accidental quota consumption rather than increasing the quota itself.

I recently used most of my 5-hour Work/Codex allowance on a relatively trivial task because the new Work chat was using an expensive model. I already know that I can manually select a less expensive model, but remembering to do that for every new Work chat is unnecessarily error-prone.

Normal Chat and Work also look and behave similarly enough when starting a conversation that it is easy to overlook which model/reasoning configuration Work has selected.

A persistent default would let users make the inexpensive model their normal Work configuration and explicitly opt into the more expensive models only when a task warrants them.

No files, tests, or entry points are named. Start by locating the macOS Work chat settings and new-chat initialization, then trace how model and reasoning selections are stored. Done means users can set Work defaults, optionally remember the last selection, and still override both values for an individual chat.