Desktop app model picker does not show gpt-5.5 even when backend reports it available

Summary

The Codex desktop app model picker does not show gpt-5.5, even though the app backend reports that model as available and the current config is already set to gpt-5.5.

This looks like a desktop UI / model-filtering issue rather than a backend availability problem.

Reproduction

1. Start the Codex desktop app.
2. Open the model picker before starting a new chat.
3. Observe that gpt-5.5 is missing from the selectable models.

Expected behavior

If the backend reports gpt-5.5 in the available model list, the desktop app should display it in the picker and allow selecting it.

Actual behavior

The desktop app picker does not show gpt-5.5, so a new chat falls back to another model (for example gpt-5.4) even when gpt-5.5 is otherwise recognized by the app runtime.

Evidence

• The app backend model/list response includes gpt-5.5.
• The app backend config/read response shows the configured model is gpt-5.5.
• The desktop picker still does not display gpt-5.5.

Additional notes

This may be related to the same class of desktop model-filtering / feature-flag issues discussed in:

• CLI-3934
• [#10873](<https://github.com/openai/codex/issues/10873>)

From local inspection, the frontend appears to apply an additional filtering step before rendering the picker, so this may be a UI-side gating/regression rather than a provider/config problem.