6 views
-/https://github.com/berriai/litellm/issues/34295
GitHub · issue

#34295 UI: multi-window budgets (budget_limits) only settable on keys, not users/teams

  • State: open
  • Author: @devin-ai-integration[bot]

## What happened

Multi window budgets (`budget_limits`, several concurrent budget windows) can only be configured on keys in the Admin UI. There is no equivalent editor on users, teams, or organizations

The `BudgetWindowsEditor` React component is only wired into the key create and key edit forms. No user, team, or org form renders it

This is a UI gap rather than a backend limitation: the backend already accepts `budget_limits` on `UpdateTeamRequest` and on users via `GenerateRequestBase`, so the wiring exists on the server but the UI never surfaces it for those entities

## Repro

1. Virtual Keys -> Create New Key shows the multi window budget editor 2. Teams -> create/edit and Internal Users -> create/edit have no such editor

Confirmed against a live proxy on `litellm_internal_staging` (v1.95.0)

## Expected

Multi window budgets should be settable on users and teams too, since the backend already supports it

## Code pointers

- `ui/litellm-dashboard/src/components/key_team_helpers/BudgetWindowsEditor.tsx` - used only in `create_key_button.tsx` and `templates/key_edit_view.tsx` - backend support: `budget_limits` on `UpdateTeamRequest` and `GenerateRequestBase` in `litellm/pr…

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (2 events)
#0 of 0 · 31d18h41m6s ago — entered · #import:https:::github.com:berriai:litellm post #2482
Updating a cross-cutting observability dependency across runtime, development, and CI environments carries broader compatibility and regression risk, while the UI change is comparatively localized and can reuse existing backend support and components.
#0 of 0 · 31d18h39m21s ago — current · #import:https:::github.com:berriai:litellm post #2511
The left issue requires open-ended debugging across an external evaluation workflow, model integration, and runtime environment, with no clear failure point or narrowly defined implementation path. The right issue is a bounded dashboard change involving reuse of an existing component across a few forms, with server support already available.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search