Problem description
The web dashboard shows all 157 models from the catalog, but only models in the active set are actually probed. The other 152 models show as "pending" forever, which confuses users into thinking the testing is broken.
Current behavior
- Dashboard shows 157 models with statuses: "up", "down", "pending"
- Only models in the active set (default: 5 models) are probed
- Models not in the active set stay "pending" indefinitely
- No visual distinction between "pending because testing is in progress" vs "pending because not in set"
Expected behavior
Users should clearly understand:
- Which models are being tested (in active set)
- Which models are not being tested (not in active set)
- Why a model shows "pending"
Suggested improvements
- Visual distinction: Add an icon or label to models not in the active set (e.g., "not in set" badge)
- Clarify "pending": Show "pending (not in set)" vs "pending (testing...)"
- Filter by set: Add a filter to show only models in the active set
- Tooltip: Hovering over "pending" shows why (e.g., "This model is not in your active router set")
Impact
- Users think the testing is broken when it's actually working as designed
- New users don't understand why some models show "pending" forever
- Confusion leads to support requests and frustration
Notes
This affects all surfaces (CLI, Docker, Desktop) since the model list rendering is shared.
Problem description
The web dashboard shows all 157 models from the catalog, but only models in the active set are actually probed. The other 152 models show as "pending" forever, which confuses users into thinking the testing is broken.
Current behavior
Expected behavior
Users should clearly understand:
Suggested improvements
Impact
Notes
This affects all surfaces (CLI, Docker, Desktop) since the model list rendering is shared.