The screenshot is created with an inprogress version, the unit is now added to the speed, e.g. 52.96/s

i think it might be helpful to have at least tok/s as the unit, or tokens/s or similar. /s is probably not enough context if we are going to add this

This is much better, and I realize I have a few more things

I think the prefill we should change to something more like

Processed ${numTokens} input tokens in ${ppTimeMs > 1000 ? `${(ppTimeMs / 1000).toFixed(2)}s` : `${ppTimeMs}ms`}

For 'predicted' I think more common language might be:

Generated ${numTokens} tokens in ${tgTimeMs > 1000 ? `${(tgTimeMs / 1000).toFixed(2)}s` : `${tgTimeMs}ms`}

Basically just making the language a bit more familiar to average users as well as moving to seconds for the time when it's greater than a 1000 milliseconds as it'll be easier to read.

Another note is the info button just looks visually a little large in the screenshot, not sure if this is just my eyesight, but if it could be more visually similar in size to the copy button next to it that would be great. I think it's mostly due to the weight (bold) of the icon

Yeah, I also noticed that thing with the buttons, I changed it to text only:

Nice this is looking great, thank you for the contribution. I'll test it on my machine today or tomorrow and pull it in

I tested it and for the info button, if it's possible to show the help text immediately or upon clicking the button, I think it would improve the user experience. Right now it's not very obvious that you have to wait a few seconds before the help text pops up.

The copy button is great.

I want to create a pull request in the next days to make the whole chat more mobile friendly, I planned there to show it on click, because on touch devices there's no tooltip at all. So I would keep it for now as it is and have the logic itself in place, and adjust it in that future pull request.

Would you want to do that all within one pull request? I can pull this in as is if it's easier for you that way, but I would prefer one pull request with all of the changes if possible. I'll be cutting a new release in the next few days.

I'm happy to leave out these changes for the next release, but can include them if they are ready

That also depends on you whether you like smaller iterations with any value or bigger but more complete ones.

I would prefer more complete in this specific instance, really appreciate all your help, it's made my life a lot easier so many thanks

Small screen UI (<768px)

Thank you so much for the contribution. I just made a few tiny tweaks.