assistant: pick the preferred model, loaded on demand

The daemon answers with whichever model llama-server has loaded and
never asks for one, so it never evicts a model another client is using.
The Assistant settings page now lists the router's models with their
state. Load saves one as the preferred model, in Prefs, and asks the
router to load it now, polling /v1/models until it reports loaded;
Prefer marks the loaded model as preferred without swapping. The swap
happens only on that click, so the daemon's rule holds.

AGENTS.md records the router's load endpoint and its two refusals, both
measured: it lives on the server root, not under /v1.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

1 files changed, 68 insertions, 0 deletions

diff --git a/desktop/modules/assistant/AssistantPage.qml b/desktop/modules/assistant/AssistantPage.qml
index e5d76c8..2e40ed4 100644
--- a/desktop/modules/assistant/AssistantPage.qml
+++ b/desktop/modules/assistant/AssistantPage.qml
@@ -162,6 +162,74 @@ Column {
 
     Heading { text: "Model" }
 
+    Small {
+        width: page.width
+        wrapMode: Text.Wrap
+        text: "The assistant answers with whichever model llama-server has loaded. Load makes one the "
+            + "preferred model and loads it now, which swaps out the loaded one for every client."
+    }
+
+    Small {
+        visible: page.asst.models.length === 0
+        text: "llama-server is not answering at " + page.asst.llm
+        color: Theme.red
+    }
+
+    Repeater {
+        model: page.asst.models
+
+        delegate: Rectangle {
+            id: modelRow
+            required property var modelData
+            readonly property bool loaded: modelData.status === "loaded"
+            readonly property bool busy: page.asst.loading === modelData.id
+            readonly property bool preferred: Prefs.assistantModel === modelData.id
+            objectName: "model:" + modelData.id
+            width: page.width
+            height: 40
+            radius: 8
+            color: loaded ? Qt.alpha(Theme.accent, 0.14) : Qt.alpha(Theme.surface, 0.35)
+            border.width: 1
+            border.color: loaded ? Qt.alpha(Theme.accent, 0.5) : "transparent"
+
+            Rectangle {
+                id: dot
+                anchors { left: parent.left; leftMargin: 12; verticalCenter: parent.verticalCenter }
+                width: 8; height: 8; radius: 4
+                color: modelRow.loaded ? Theme.green : modelRow.busy ? Theme.yellow : Theme.overlay
+            }
+            Text {
+                id: modelName
+                anchors { left: dot.right; leftMargin: 10; verticalCenter: parent.verticalCenter }
+                text: modelRow.modelData.id
+                color: Theme.text
+                font { family: Theme.fontFamily; pixelSize: Theme.fontSize - 2; bold: modelRow.loaded }
+            }
+            Small {
+                anchors { left: modelName.right; leftMargin: 10; verticalCenter: parent.verticalCenter }
+                text: [modelRow.busy ? "loading…" : modelRow.modelData.status,
+                       modelRow.preferred ? "preferred" : ""].filter(s => s).join("  ·  ")
+                color: modelRow.preferred ? Theme.accent : Theme.subtext
+            }
+            Button {
+                anchors { right: parent.right; rightMargin: 6; verticalCenter: parent.verticalCenter }
+                text: modelRow.loaded ? (modelRow.preferred ? "Loaded" : "Prefer") : "Load"
+                enabled: page.asst.loading === "" && !(modelRow.loaded && modelRow.preferred)
+                // A loaded model only becomes the preferred one: nothing to swap.
+                onClicked: modelRow.loaded ? Prefs.setAssistantModel(modelRow.modelData.id)
+                                           : page.asst.loadModel(modelRow.modelData.id)
+            }
+        }
+    }
+
+    Small {
+        width: page.width
+        visible: page.asst.modelError !== ""
+        wrapMode: Text.Wrap
+        text: page.asst.modelError
+        color: Theme.red
+    }
+
     Labelled {
         objectName: "key:llm"
         label: "Server"