---
title: "Comment by Kartik N V J K on \"The dashboard is green and one of your models is not on it\" | Hashnode"
description: "The line about an absent metric looking exactly like a low one is the trap, because auto-instrumentation is sold as the thing you stop thinking about, which is"
url: "https://hashnode.com/posts/the-dashboard-is-green-and-one-of-your-models-is-not-on-it/6a8158f7ad2d365d301c37ab/comment/6a831f5ce1cd816acf9acde6"
---

[Discussion](https://hashnode.com/posts/the-dashboard-is-green-and-one-of-your-models-is-not-on-it/6a8158f7ad2d365d301c37ab)

Comment [The dashboard is green and one of your models is not on it](https://hashnode.com/posts/the-dashboard-is-green-and-one-of-your-models-is-not-on-it/6a8158f7ad2d365d301c37ab)

[Kartik N V J K](https://hashnode.com/@kartik-nvj)

[Aug 17](https://hashnode.com/posts/the-dashboard-is-green-and-one-of-your-models-is-not-on-it/6a8158f7ad2d365d301c37ab/comment/6a831f5ce1cd816acf9acde6)

The line about an absent metric looking exactly like a low one is the trap, because auto-instrumentation is sold as the thing you stop thinking about, which is precisely why nobody audits its gaps. I run a synthetic call with known token counts through each model path and fail the check if the emitted usage does not match, which surfaces the silently-dropped path fast. Do you assert expected-vs-emitted token counts anywhere, or catch these by eyeballing the cost chart?
