model temperature settings

packages/chat-client/src/components/SettingsView.jsx

@@ -241,6 +241,7 @@ function ServiceHealth() {
 function ModelsSection({ onNavigate }) {
   const { models, selectedModel, setSelectedModel } = useModels();
   const [selectedInfo, setSelectedInfo] = useState(null);
+  const {settings, saveSetting, saving} = useSettings();
 
   // Sync info panel when selection changes
   useEffect(() => {
@@ -255,6 +256,19 @@ function ModelsSection({ onNavigate }) {
         description="Path to folder containing .gguf files"
         action={<ModelsFolderSetting />}
       />
+      <SettingsRow
+        label="Temperature"
+        description="Response randomness — lower is more focused, higher is more creative (0–2)"
+        action={
+          <NumberSetting
+            label=""
+            value={settings?.temperature}
+            min={0} max={2} step={0.05}
+            onSave={val => saveSetting('temperature', val)}
+            saving={saving}
+          />
+        }
+      />
       <SettingsRow
         label="Active Model"
         description="Model used for inference"

packages/orchestration-service/src/chat/index.js

@@ -126,7 +126,7 @@ async function getRelevantEntities(userMessage) {
 }
 
 async function chat(externalId, userMessage, options = {}) {
-  const { recentEpisodeLimit, semanticLimit, scoreThreshold } =
+  const { recentEpisodeLimit, semanticLimit, scoreThreshold, temperature} =
     appSettings.load();
   // 1. Resolve or create session
   let session = await memory.getSessionByExternalId(externalId);
@@ -187,7 +187,7 @@ async function chat(externalId, userMessage, options = {}) {
   );
 
   // 5. Run inference
-  const result = await inference.complete(prompt, options);
+  const result = await inference.complete(prompt, {...options, temperature});
 
   // 6. Write episode back to memory
   memory
@@ -217,7 +217,7 @@ async function chat(externalId, userMessage, options = {}) {
 
 async function chatStream(externalId, userMessage, onChunk, options = {}) {
   try {
-    const { recentEpisodeLimit, semanticLimit, scoreThreshold } = appSettings.load();
+    const { recentEpisodeLimit, semanticLimit, scoreThreshold, temperature } = appSettings.load();
     let session = await memory.getSessionByExternalId(externalId);
     if (!session) session = await memory.createSession(externalId);
 
@@ -270,7 +270,7 @@ async function chatStream(externalId, userMessage, onChunk, options = {}) {
       entities,
       userMessage,
     );
-    const res = await inference.completeStream(prompt, options);
+    const res = await inference.completeStream(prompt, {...options, temperature});
 
     let fullText = "";
     let model = "";

packages/orchestration-service/src/config/settings.js

@@ -8,7 +8,8 @@ const DEFAULTS = {
   recentEpisodeLimit:   ORCHESTRATION.RECENT_EPISODE_LIMIT,
   semanticLimit:        ORCHESTRATION.SEMANTIC_LIMIT,
   scoreThreshold:       ORCHESTRATION.SCORE_THRESHOLD,
-  modelsFolderPath:     getEnv('MODELS_MANIFEST_PATH', '/mnt/nexus-models')
+  modelsFolderPath:     getEnv('MODELS_MANIFEST_PATH', '/mnt/nexus-models'),
+  temperature:          ORCHESTRATION.TEMPERATURE
 };
 
 function load() {

packages/shared/src/config/constants.js

@@ -25,6 +25,7 @@ const ORCHESTRATION = {
     RECENT_EPISODE_LIMIT:   5,
     SEMANTIC_LIMIT:         5,
     SCORE_THRESHOLD:        0.75,
+    TEMPERATURE:            0.7,
     CORS_ORIGIN:            'http://localhost:5173',
     SYSTEM_PROMPT:          `You are a helpful, context-aware AI assistant. You have access to memories of past conversations with the user. Use them to provide consistent, personalised responses.`
 }