[FEATURE] Small, possibly local, model for during big tasks and after usage limit

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

When the main chat is running tasks and gathering data, nothing else can be done really and it's a waiting game. also, when the usage limit is reached, nothing can be done until it iss reset without using api credits. This also means planning and other small prep steps cant happen.

Proposed Solution

With the rise of NPU's in processors, along side more AI focused discrete GPUs, it seems like a waste to not leverage them in some capacity.

I'd like it, so that when the main chat is running or usage limit is reached, it also has access to a small local model. This would be most useful in a new window/seperate chat, but still connected to the main session. This model wouldn't be able to edit code, or plan full projects, or write an essay ect. It would probably be quite slow. but the ability to add things to a to-do list, for instance, would be very help full. and then when the main chat is ready or the usage is reset it will also have access to this to-do list or notes and you can then direct it to pull information from there.

It can remain seperate from the main chat by having the main chat, in the background, keep an updated, very simplified running summary of the current state of the project, which the simple chat has read-only access to and then the simple change can write things to a to-do list ect. that the main chat has read-only access to.

Alternative Solutions

i cannot think of any, outside of pen and paper

Priority

Medium - Would be very helpful

Feature Category

Other

Use Case Example

1. i've set it off on a long set of tasks, its adding code, its writing and running tests, its reading the existing code base ect. its going to take over 30min
2. 5 min in, I think of the next thing i want it to do.
3. i turn to the second chat, the local model. i describe what i want to do next.
4. it thinks about it, it uses some context from the main session maybe, it adds the new task to the to-do list.
5. 10 min aftert that i add a second, maybe a third.
6. 30 minutes later, when i would have forgotten what i wanted to do next normally, the to-do is there ready and waiting and i can just get the main agent to reference that and continue.
7. i have to leave for a couple of hours
8. 2 hours later i still have what i wrote down before and its sitting there ready to be fed into the main chat.

Additional Context

_No response_