fix(ui): honor model view scope

[lzhan011]

TLDR

Problem this solves:

• All Available Models still used the selected team filter
• Scope changes retained pagination from the previous dataset
• The current UI lint budget exceeded its prefer-screen-queries limit

How it solves it:

• Omit teamId when viewing all available models
• Reset pagination whenever the view scope changes
• Restore the selected team scope when switching back
• Replace render-result queries in the models page tests with screen queries

User Flow

Before: an admin cannot reliably leave the selected team's model scope

1. They open https://litellm-domain/ui/models-and-endpoints
2. They select Team "Engineering" and go to page 2
3. They select View "All Available Models"
4. The model request still includes teamId=team-1 and page=2
5. The table remains team-scoped or appears empty

After: the same view switch loads all available models from the first page

1. They open https://litellm-domain/ui/models-and-endpoints
2. They select Team "Engineering" and go to page 2
3. They select View "All Available Models"
4. The model request omits teamId and uses page=1
5. The table shows all available models
6. Switching back restores teamId=team-1 from page 1

Relevant issues

Discovered through a static query-state audit and the PR lint check

Linear ticket

Pre-Submission checklist

• I have added meaningful tests
• The focused component test files pass locally
• The complete UI lint budget check passes locally
• My PR passes all required CI/CD checks
• My PR's changes are limited to the models-and-endpoints UI
• I have received a Greptile Confidence Score of at least 4/5 on the latest head

Screenshots / Proof of Fix

Before (bad55da)

1. Select Engineering in the team control and advance to page 2
2. Select All Available Models in the view control
3. The interaction regression receives teamId: "team-1" and page: 2
4. The new test fails because the expected unscoped first-page query never occurs
5. UI Lint reports testing-library/prefer-screen-queries: 21 | max: 18

After (e3ed5fd)

1. Repeat the same team, pagination, and view selections
2. The interaction regression receives teamId: undefined and page: 1
3. Switch back to Current Team Models
4. The next request receives teamId: "team-1" from page 1
5. Both focused component files pass: 40 tests
6. The full lint budget reports testing-library/prefer-screen-queries: 18 | max: 18

Type

Bug Fix

Caveats (if any)

CI baseline

• OSV currently flags pypdf 6.15.0 and tornado 6.5.7 from the staging uv.lock
• Fork PRs are prohibited from modifying uv.lock by the repository dependency guard
• Canonical dependency PR fix(deps): raise the tornado and pypdf floors for six new advisories #39188 is tracking the staging security update

Final Attestation

• The tests cover scope changes from page 2 and the restore path
• The complete lint budget check passes at its configured limits

[greptile-apps]

Greptile Summary

The PR corrects model query scoping and ensures scope changes begin on the first page.

• Omits the selected team ID while viewing all available models and restores it for the current-team view.
• Resets pagination whenever the model view changes.
• Adds interaction coverage for switching from a team-scoped second page to the unscoped first page.
• Updates page tests to use global screen queries.

Confidence Score: 5/5

The PR appears safe to merge.

The previously reported pagination issue is fixed: changing model scope now resets pagination while the query derives the new page and team scope together, and no blocking failure remains.

Important Files Changed

Filename	Overview
ui/litellm-dashboard/src/app/(dashboard)/models-and-endpoints/components/AllModelsTab.tsx	The scope-dependent team query and pagination reset correctly address the previously reported stale-page behavior.
ui/litellm-dashboard/src/app/(dashboard)/models-and-endpoints/components/AllModelsTab.test.tsx	The added interaction test verifies that switching from team page two to all models requests the unscoped first page and restores the selected team afterward.
ui/litellm-dashboard/src/app/(dashboard)/models-and-endpoints/page.test.tsx	The test-query changes are a semantics-preserving cleanup from render-scoped queries to screen.

_{Reviews (3): Last reviewed commit: "test(ui): use screen queries in models p..." | Re-trigger Greptile}

[lzhan011]

@greptileai please re-review the pagination reset and patched browserslist lockfile at the latest head commit

[codspeed-hq]

Merging this PR will not alter performance

✅ 31 untouched benchmarks

---

_{Comparing lzhan011:litellm_fix_model_view_scope (9afb744) with litellm_internal_staging (8469003)}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[lzhan011]

@greptileai please re-review the model scope fix and screen-query lint cleanup at the latest head commit

[lzhan011]

OSV failure is tracked in #39217 and comes from unchanged staging locks; this PR only changes model scope UI code and tests