qol(virtual-quota)

[ivanarifin]

Context

1. Reduce cool down to 10s: the rate limit errors sometime is fixed straight away by refreshing the request, the cool down is still needed but the 10 minutes cool down is too long
2. Show which profile currently active in the model selector: this gives more info to the user on which profile is currently being used
3. Supports image by default: this will help with model that supports image but the model is still not recognized yet by the virtual quota fallback provider
4. Change default context window to 1M so it won't show red bar on the context used

Implementation

Screenshots

before	after

How to Test

Use the provider

Get in Touch

Ivanarifin

[changeset-bot]

⚠️ No Changeset found

Latest commit: 3e858dc

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[ivanarifin]

hey @chrarnoldus @mcowger need your thought and review on this PR. Thanks!

[mcowger]

Reduce cool down to 10s: the rate limit errors sometime is fixed straight away by refreshing the request, the cool down is still needed but the 10 minutes cool down is too long

Seems reasonable.

Show which profile currently active in the model selector: this gives more info to the user on which profile is currently being used

Nice.

Supports image by default: this will help with model that supports image but the model is still not recognized yet by the virtual quota fallback provider

Not sure I'm a fan of this. Wouldn't it result in a user experiencing confusion?

Change default context window to 1M so it won't show red bar on the context used

Wasn't this fixed with your last round of changes that updated the information as the model switched?

[ivanarifin]

Not sure I'm a fan of this. Wouldn't it result in a user experiencing confusion?

okay it actually make sense, giving the user 404 error is not a good experience, reverted it

Wasn't this fixed with your last round of changes that updated the information as the model switched?

this is actually just a UI "improvement", at least that's what i think. the current context window is 1 by default. My case, for example:
Using the virtual quota for code mode, so from architect mode using a different profile, then switch to code mode, before the virtual quota recognize the model, the context window bar on top will always change to red, as if the context window is full, this change in the PR just to make it not seem like it's full, it's not critical

The critical one for me is actually the cooldown, my case:
I have 2 profiles of gemini cli,i got free tier rate limit, then switch to the other profile, then got the same error, then all profiles are under cooldown. But when i tried to use the profile again, not from virtual quota, the gemini cli rate limit is fixed. So i think the rate limit needed is the one that the user set

[ivanarifin]

so any more feedback on this @mcowger @chrarnoldus ? is it possible for this PR to be merged?