diff --git a/src-tauri/src/lib.rs b/src-tauri/src/lib.rs
index 1baef020b..d5e607ff6 100644
--- a/src-tauri/src/lib.rs
+++ b/src-tauri/src/lib.rs
@@ -8,8 +8,7 @@ use core::{
 };
 use reqwest::Client;
 use std::{collections::HashMap, sync::Arc};
-use tauri::{Emitter, Manager};
-
+use tauri::{Emitter, Manager, RunEvent};
 use tokio::sync::Mutex;
 
 #[cfg_attr(mobile, tauri::mobile_entry_point)]
@@ -22,7 +21,8 @@ pub fn run() {
           // when defining deep link schemes at runtime, you must also check `argv` here
         }));
     }
-    builder
+
+    let app = builder
         .plugin(tauri_plugin_os::init())
         .plugin(tauri_plugin_deep_link::init())
         .plugin(tauri_plugin_dialog::init())
@@ -143,12 +143,46 @@ pub fn run() {
                         cleanup_processes(state).await;
                     });
                 }
+
                 let client = Client::new();
                 let url = "http://127.0.0.1:39291/processManager/destroy";
                 let _ = client.delete(url).send();
             }
             _ => {}
         })
-        .run(tauri::generate_context!())
+        .build(tauri::generate_context!())
         .expect("error while running tauri application");
+
+    // Handle app lifecycle events
+    app.run(|app, event| match event {
+        RunEvent::Exit => {
+            // This is called when the app is actually exiting (e.g., macOS dock quit)
+            // We can't prevent this, so run cleanup quickly
+            let app_handle = app.clone();
+            tokio::task::block_in_place(|| {
+                tauri::async_runtime::block_on(async {
+                    let state = app_handle.state::<AppState>();
+
+                    // Hide window immediately
+                    if let Some(window) = app_handle.get_webview_window("main") {
+                        let _ = window.hide();
+                        let _ = window.emit("kill-mcp-servers", ());
+                    }
+
+                    // Quick cleanup with shorter timeout
+                    cleanup_processes(state).await;
+
+                    // Stop HTTP server with shorter timeout
+                    let client = Client::new();
+                    let url = "http://127.0.0.1:39291/processManager/destroy";
+                    let _ = tokio::time::timeout(
+                        tokio::time::Duration::from_secs(2),
+                        client.delete(url).send(),
+                    )
+                    .await;
+                });
+            });
+        }
+        _ => {}
+    });
 }
diff --git a/web-app/src/hooks/__tests__/useThreads.test.ts b/web-app/src/hooks/__tests__/useThreads.test.ts
index 3c7915206..12caa5b59 100644
--- a/web-app/src/hooks/__tests__/useThreads.test.ts
+++ b/web-app/src/hooks/__tests__/useThreads.test.ts
@@ -88,7 +88,7 @@ describe('useThreads', () => {
     })
 
     expect(Object.keys(result.current.threads)).toHaveLength(2)
-    expect(result.current.threads['thread1'].model.id).toEqual('thread1:free')
+    expect(result.current.threads['thread1'].model.id).toEqual('thread1/free')
     expect(result.current.threads['thread1'].model.provider).toEqual('llamacpp')
     expect(result.current.threads['thread2'].model.id).toEqual('thread2/test')
     expect(result.current.threads['thread2'].model.provider).toEqual('llamacpp')
diff --git a/web-app/src/hooks/useThreads.ts b/web-app/src/hooks/useThreads.ts
index 140972720..83bd320c8 100644
--- a/web-app/src/hooks/useThreads.ts
+++ b/web-app/src/hooks/useThreads.ts
@@ -44,9 +44,11 @@ export const useThreads = create<ThreadState>()((set, get) => ({
                   'llamacpp'
                 ),
                 // Cortex migration: take first two parts of the ID (the last is file name which is not needed)
-                id: !thread.model?.id.endsWith(':free')
-                  ? thread.model?.id.split(':').slice(0, 2).join(sep())
-                  : thread.model?.id,
+                id:
+                  thread.model.provider === 'llama.cpp' ||
+                  thread.model.provider === 'llamacpp'
+                    ? thread.model?.id.split(':').slice(0, 2).join(sep())
+                    : thread.model?.id,
               }
             : undefined,
         }
diff --git a/web-app/src/routes/settings/providers/$providerName.tsx b/web-app/src/routes/settings/providers/$providerName.tsx
index e3dc7af6c..326500273 100644
--- a/web-app/src/routes/settings/providers/$providerName.tsx
+++ b/web-app/src/routes/settings/providers/$providerName.tsx
@@ -133,22 +133,6 @@ function ProviderDetail() {
   // Note: settingsChanged event is now handled globally in GlobalEventHandler
   // This ensures all screens receive the event intermediately
 
-  // Auto-refresh models for non-predefined providers
-  useEffect(() => {
-    if (
-      provider &&
-      provider.provider !== 'llamacpp' &&
-      !predefinedProviders.some((p) => p.provider === provider.provider) &&
-      provider.base_url
-    ) {
-      // Auto-refresh models every 10 seconds for remote providers
-      const intervalId = setInterval(() => {
-        handleRefreshModels()
-      }, 10000)
-      return () => clearInterval(intervalId)
-    }
-  }, [provider])
-
   const handleJoyrideCallback = (data: CallBackProps) => {
     const { status } = data