[BUG] IO Intelligence model picker reverts to previous selection

[hannesrudolph]

Type

Bug

Problem

When editing an API profile that uses IO Intelligence, changing the model does not stick. The selection snaps back to the previously selected model (e.g., "openai/gpt-oss-120b").

Context

Occurs in Settings > Providers while editing an API profile for the IO Intelligence provider. The IO Intelligence API key is set. The revert happens immediately upon selecting a different model from the dropdown (before saving).

Reproduction

1. Open Settings > Providers
2. Select an API profile configured for IO Intelligence (ensure IO Intelligence API key is set)
3. Open the Model dropdown
4. Select a different model

Expected result: The newly selected model remains selected
Actual result: The selection immediately reverts to the previous model (e.g., "openai/gpt-oss-120b")

[roomote]

I'm exploring a potential fix for the IO Intelligence model picker issue and will open a PR to share progress.

[roomote]

Opened PR #7993 with my attempt at addressing this issue: #7993

The fix prevents the ModelPicker component from reinitializing when the model field has already been set, which was causing the IO Intelligence model selection to revert.

Note: There's currently a Windows test failure in CI that appears to be unrelated to this change, as the same test passes on Ubuntu. Feedback welcome!

[hannesrudolph]

Cannot repro