You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Cursor's Today / Yesterday / Last 30 Days tiles (and the spend-tile hover) treat the whole usage export as one pool. Grok Bot rows in that export use grok-bot-* slugs (grok-bot-default, grok-bot-cua, grok-bot-automation). The pricing supplement aliases those slugs to grok-4.6, the same family as IDE cursor-grok-* rows, so the hover ranking collapses both products into one Grok family.
The existing Grok Bot widget from #1132 / #1134 is only the separate weekly allowance percent. It does not say how much of the estimated spend tiles came from Grok Bot versus Cursor Grok (or the rest of Cursor).
Proposed Solution
Keep the combined spend tiles as the Total Spend contribution (one Cursor total, no double-count). When a period includes both products, have the spend-tile hover list Cursor and Grok Bot first instead of mixing them under grok-4.6.
Classification can stay on the export's Model column: slugs starting with grok-bot- are Grok Bot; everything else stays Cursor. Periods that only used one product can keep today's per-model family ranking.
No extra Cursor endpoints are required for this. The weekly Grok Bot meter can stay as it is.
Alternatives Considered
Separate Grok Bot spend tiles (another Today / Yesterday / Last 30 Days set). Clearer at a glance, but six more metrics, new layout defaults, and a second "Grok Bot" title next to the weekly percent meter. Heavier than this repo usually wants for a first step.
A second provider card. Grok Bot already bills through the Cursor login and Cursor & Grok: Add Grok Bot usage #1132 placed its allowance under Cursor. A new provider would duplicate auth and invite double-counting in Total Spend.
Filter GetFilteredUsageEvents with clientType=sand. That path is for the weekly meter and is not a clean product split on the export used for spend tiles. The CSV model slugs already name Grok Bot.
Problem
Cursor's Today / Yesterday / Last 30 Days tiles (and the spend-tile hover) treat the whole usage export as one pool. Grok Bot rows in that export use
grok-bot-*slugs (grok-bot-default,grok-bot-cua,grok-bot-automation). The pricing supplement aliases those slugs togrok-4.6, the same family as IDEcursor-grok-*rows, so the hover ranking collapses both products into one Grok family.The existing Grok Bot widget from #1132 / #1134 is only the separate weekly allowance percent. It does not say how much of the estimated spend tiles came from Grok Bot versus Cursor Grok (or the rest of Cursor).
Proposed Solution
Keep the combined spend tiles as the Total Spend contribution (one Cursor total, no double-count). When a period includes both products, have the spend-tile hover list Cursor and Grok Bot first instead of mixing them under
grok-4.6.Classification can stay on the export's
Modelcolumn: slugs starting withgrok-bot-are Grok Bot; everything else stays Cursor. Periods that only used one product can keep today's per-model family ranking.No extra Cursor endpoints are required for this. The weekly Grok Bot meter can stay as it is.
Alternatives Considered
GetFilteredUsageEventswithclientType=sand. That path is for the weekly meter and is not a clean product split on the export used for spend tiles. The CSV model slugs already name Grok Bot.Additional Context
Related, not a duplicate:
Costcolumn), a different axis from product.docs/providers/cursor.mddocuments the weekly meter; spend history is still described as one export total.Happy to implement if a maintainer approves this and assigns it.