Weird behaviour of model selector

[Konstantinov-Innokentii]

1. If user selects a model on existing conversation – we don't show the user override chip
2. When model is changed via api key selection we choose the "best" model for this key. if the selected model matches agent/org model it will anyway be shown as "user override"

[London-Cat]

📊 Reputation Summary

User	Rep	Pull Requests	Activity	Assigned	Core Reactions
Konstantinov-Innokentii	⚡ 1868	86✅ 1🔄 13❌	55 issues, 50 comments	7	—
Dev0907	⚡ 3	0✅ 1🔄 0❌	0 issues, 2 comments	0	—
1164148-7-eng	⚡ -5	0✅ 5🔄 2❌	0 issues, 24 comments	0	—

---

How is the score calculated? Read about it in the Reputation Bot repository 🤖

[Dev0907]

Hi @Konstantinov-Innokentii Konstantinov-Innokentii, I’d like to work on this issue.

From my understanding, the inconsistency comes from how we determine and display the model override state.

Proposed approach:

• Introduce a clear distinction between user-selected models and auto-selected models (via API key)
• Ensure the override chip is shown:
  • when the user explicitly selects a model
  • when the selected model differs from the default (agent/org)
• Ensure the override chip is NOT shown:
  • when the model is auto-selected and matches the default model

Implementation plan:

• Update the model selection logic to track whether the selection is user-driven
• Refactor the override condition to use both:
  • isUserSelected
  • model comparison with default
• Test across both manual selection and API key flows to ensure consistent behavior

Please let me know if this approach looks good, and I can proceed with the implementation.

[1164148-7-eng]

/attempt #3491

I've claimed this issue. Working on it now! 🚀