When the model is responding, switching to another conversation causes the response to be written in the new conversation

[kazu0123]

When the model is responding, switching to another conversation causes the response to be partially written in the new conversation. This behavior disrupts the consistency of conversations.

Steps to Reproduce

1. Open the Enchanted app.
2. Ask a question in Conversation A.
3. Wait for the response in Conversation A to complete.
4. Start a new conversation and ask a question (referred to as Conversation B).
5. While the model is responding in Conversation B, switch back to Conversation A.
6. Part of the response from Conversation B appears in Conversation A.

Expected Behavior

• The model's response should only be written in the conversation where the question was asked (Conversation B).
• Other conversations (e.g., Conversation A) should remain unaffected.

Actual Behavior

• The model's response is partially written in Conversation A.

Environment

• Enchanted Version 1.7.2 (50)
• macOS Sequoia Version 15.1.1 (24B2091)
• ollama version is 0.4.3

Video (Steps 4 onward)