@@ -155,7 +155,7 @@ A multimodal AI system for the Pepper robot powered by advanced speech-to-speech
 - **Real-time Voice Chat** - Natural conversations powered by speech-to-speech (S2S) AI models:
   - **OpenAI** - gpt-realtime, gpt-realtime-mini, gpt-4o-realtime-preview, gpt-4o-mini-realtime-preview
   - **Azure OpenAI** - Same models with network-level isolation and customer-managed encryption for enterprise compliance
-  - **x.ai** - Grok Voice Agent API with native web/X search
+  - **x.ai** - `grok-voice-think-fast-1.0` (default) and `grok-voice-fast-1.0` Voice Agent API with native web/X search
   - **Google** - Gemini Live API with native audio model and 30 voices
 - **Audio Input Modes** - Two optimized modes for voice capture:
   - **Direct Audio** - Low-latency streaming with native voice activity detection
@@ -476,12 +476,18 @@ This allows the Android app to automatically start the face recognition server v
 #### Option 3: x.ai Grok Voice Agent API (Alternative with Native Search)
 1. Go to [x.ai](https://x.ai/)
 2. Create an API key
-3. Uses the Grok Voice Agent API (speech-to-speech model, OpenAI Realtime API compatible)
+3. Uses x.ai Voice Agent models (speech-to-speech, OpenAI Realtime API compatible)
+4. Add the key to `local.properties`:
+   ```properties
+   XAI_API_KEY=your_xai_api_key_here
+   ```
 
 **Unique Features:**
+- **Model Selection**: Defaults to `grok-voice-think-fast-1.0`; `grok-voice-fast-1.0` remains selectable in settings
+- **Explicit Realtime Model URL**: Uses `wss://api.x.ai/v1/realtime?model=<selected-model>`
 - **Native Web Search**: Built-in web search without Tavily API (configurable in settings)
 - **Native X Search**: Search posts on X/Twitter in real-time (configurable in settings)
-- **5 Distinct Voices**: Ara, Rex, Sal, Eve, Leo
+- **5 Distinct Voices**: `eve`, `ara`, `rex`, `sal`, `leo`
 - **100+ Languages**: Multilingual support out of the box
 - **Note**: Vision analysis requires `GROQ_API_KEY` (Grok Voice Agent API doesn't support images)
 
@@ -639,7 +645,7 @@ You can switch between different AI providers in the settings:
 
 - **OpenAI Direct** (Realtime API): Supports all four models (`gpt-realtime`, `gpt-realtime-mini`, `gpt-4o-realtime-preview`, `gpt-4o-mini-realtime-preview`) directly from OpenAI with 10 voices
 - **Azure OpenAI** (Realtime API): Supports all four models with your Azure deployment, network-level isolation, and customer-managed encryption for enterprise compliance
-- **x.ai Grok** (Grok Voice Agent API): Speech-to-speech voice model with native web/X search capabilities and 5 unique voices
+- **x.ai Grok** (Grok Voice Agent API): `grok-voice-think-fast-1.0` by default, with `grok-voice-fast-1.0` also selectable; native web/X search capabilities, configurable server VAD, and 5 voices (`eve`, `ara`, `rex`, `sal`, `leo`)
 - **Google Gemini** (Live API): Native audio model with end-to-end speech, configurable VAD, thinking budget, and 30 voices
 
 **Note**: Changing the API provider automatically updates available models and voices. Model/voice changes restart the session automatically.
@@ -1440,7 +1446,7 @@ Robot: "Happy Birthday! 🎂"
 ```
 
 **Key Flow Characteristics:**
-- **Model Flexibility**: Supports OpenAI (gpt-4o-realtime, gpt-4o-mini-realtime) and x.ai Grok Voice Agent
+- **Model Flexibility**: Supports OpenAI Realtime models, Google Gemini Live models, and x.ai `grok-voice-think-fast-1.0` / `grok-voice-fast-1.0`
 - **Dual Audio Input**: Realtime API (simple) or Azure Speech (dialect quality)
 - **Server-side VAD**: Realtime API handles turn detection automatically
 - **Conditional Tool Calls**: AI decides when tools are needed (not every response uses tools)
@@ -1664,6 +1670,7 @@ app/src/
 - Check your OpenAI, Azure OpenAI, x.ai, or Google API key and endpoint
 - Verify internet connectivity
 - For Azure: Ensure the deployment name matches your Azure setup
+- For x.ai: Ensure `XAI_API_KEY` is set and the selected model is `grok-voice-think-fast-1.0` or `grok-voice-fast-1.0`
 
 #### "Vision analysis not available"
 - Vision works automatically with `gpt-realtime` and `gpt-realtime-mini` models (built-in)
@@ -1762,7 +1769,7 @@ This project is licensed under the MIT License - see the [LICENSE](LICENSE) file
 ## 🙏 Acknowledgments
 
 - **OpenAI** - Realtime API
-- **x.ai** - Grok Voice Agent API
+- **x.ai** - Grok Voice Agent API (`grok-voice-think-fast-1.0`, `grok-voice-fast-1.0`)
 - **SoftBank Robotics** - Pepper robot platform
 - **Microsoft** - Azure Speech and OpenAI services
 - **Groq** - Alternative vision analysis provider
 
@@ -26,6 +26,7 @@ class RobotFocusManager(private val activity: ChatActivity) {
     var qiContext: Any? = null
         private set
     private var listener: Listener? = null
+    private var lastFocusRefusedReason: String? = null
 
     interface Listener {
         fun onRobotReady(robotContext: Any?)
@@ -51,6 +52,10 @@ class RobotFocusManager(private val activity: ChatActivity) {
                 override fun onRobotFocusLost() {
                     handleRobotFocusLost()
                 }
+
+                override fun onRobotFocusRefused(reason: String) {
+                    handleRobotFocusRefused(reason)
+                }
             })
 
             startFocusTimeout()
@@ -72,10 +77,16 @@ class RobotFocusManager(private val activity: ChatActivity) {
             if (!isFocusAvailable) {
                 Log.e(TAG, "🤖 DIAGNOSTIC: TIMEOUT - No robot focus response after 30 seconds!")
                 Log.e(TAG, "🤖 DIAGNOSTIC: Mode: ${robotController.getModeName()}")
+                lastFocusRefusedReason?.let {
+                    Log.e(TAG, "🤖 DIAGNOSTIC: Last focus-refused reason reported by QiSDK: $it")
+                }
                 if (robotController.isRobotHardwareAvailable()) {
                     Log.e(TAG, "🤖 DIAGNOSTIC: Check: 1) Robot is awake, 2) Robot is enabled, 3) QiSDK service is running")
                 }
-                listener?.onRobotInitializationFailed(activity.getString(R.string.robot_initialization_timeout))
+                val message = lastFocusRefusedReason?.let { reason ->
+                    activity.getString(R.string.robot_initialization_timeout) + " (reason: $reason)"
+                } ?: activity.getString(R.string.robot_initialization_timeout)
+                listener?.onRobotInitializationFailed(message)
             }
         }
         focusTimeoutHandler.postDelayed(focusTimeoutRunnable!!, FOCUS_TIMEOUT_MS)
@@ -108,5 +119,13 @@ class RobotFocusManager(private val activity: ChatActivity) {
 
         listener?.onRobotFocusLost()
     }
+
+    private fun handleRobotFocusRefused(reason: String) {
+        // Cache the latest refusal reason so the timeout path can surface it in the UI.
+        // We do NOT fail fast here because QiSDK sometimes recovers after a refusal
+        // and grants focus on a subsequent attempt within the timeout window.
+        Log.e(TAG, "🤖 Robot focus refused by QiSDK: $reason")
+        lastFocusRefusedReason = reason
+    }
 }
 
@@ -6,6 +6,8 @@ import android.util.Log
 import dagger.hilt.android.qualifiers.ApplicationContext
 import ch.fhnw.pepper_realtime.R
 import ch.fhnw.pepper_realtime.network.RealtimeApiProvider
+import ch.fhnw.pepper_realtime.network.XAI_LEGACY_MODEL_LABEL
+import ch.fhnw.pepper_realtime.network.XAI_THINK_FAST_MODEL
 import ch.fhnw.pepper_realtime.tools.ToolRegistry
 import ch.fhnw.pepper_realtime.ui.settings.LanguageOption
 import javax.inject.Inject
@@ -35,28 +37,38 @@ class SettingsRepository @Inject constructor(
     var model: String
         get() {
             val defaultModel = when (apiProviderEnum) {
-                RealtimeApiProvider.XAI -> "Grok Voice Agent"
+                RealtimeApiProvider.XAI -> XAI_THINK_FAST_MODEL
                 RealtimeApiProvider.GOOGLE_GEMINI -> "models/gemini-3.1-flash-live-preview"
                 else -> context.getString(R.string.openai_default_model)
             }
             val key = "${KEY_MODEL}_${apiProvider}"
-            return settings.getString(key, null)
+            val savedModel = settings.getString(key, null)
                 ?: settings.getString(KEY_MODEL, defaultModel)
                 ?: defaultModel
+            return if (apiProviderEnum == RealtimeApiProvider.XAI && savedModel == XAI_LEGACY_MODEL_LABEL) {
+                XAI_THINK_FAST_MODEL
+            } else {
+                savedModel
+            }
         }
         set(value) = settings.edit().putString("${KEY_MODEL}_${apiProvider}", value).apply()
 
     var voice: String
         get() {
             val defaultVoice = when (apiProviderEnum) {
-                RealtimeApiProvider.XAI -> "Ara"
+                RealtimeApiProvider.XAI -> "ara"
                 RealtimeApiProvider.GOOGLE_GEMINI -> "Puck"
                 else -> "ash"
             }
             val key = "${KEY_VOICE}_${apiProvider}"
-            return settings.getString(key, null)
+            val savedVoice = settings.getString(key, null)
                 ?: settings.getString(KEY_VOICE, defaultVoice)
                 ?: defaultVoice
+            return if (apiProviderEnum == RealtimeApiProvider.XAI) {
+                savedVoice.lowercase()
+            } else {
+                savedVoice
+            }
         }
         set(value) = settings.edit().putString("${KEY_VOICE}_${apiProvider}", value).apply()
 
 
@@ -2,6 +2,10 @@ package ch.fhnw.pepper_realtime.network
 
 import android.util.Log
 
+const val XAI_THINK_FAST_MODEL = "grok-voice-think-fast-1.0"
+const val XAI_FAST_MODEL = "grok-voice-fast-1.0"
+const val XAI_LEGACY_MODEL_LABEL = "Grok Voice Agent"
+
 /**
  * Enumeration of available Realtime API providers
  * Supports Azure OpenAI, direct OpenAI, and x.ai connections
@@ -12,7 +16,7 @@ enum class RealtimeApiProvider(
 ) {
     AZURE_OPENAI("Azure OpenAI", "gpt-4o-realtime-preview"),
     OPENAI_DIRECT("OpenAI Direct", "gpt-realtime-1.5"),
-    XAI("x.ai Grok", "Grok Voice Agent"),
+    XAI("x.ai Grok", XAI_THINK_FAST_MODEL),
     // Google Live API requires models/ prefix for BidiGenerateContent
     GOOGLE_GEMINI("Google Gemini", "models/gemini-3.1-flash-live-preview");
 
@@ -27,7 +31,12 @@ enum class RealtimeApiProvider(
      * @return WebSocket URL
      */
     fun getWebSocketUrl(azureEndpoint: String?, model: String?, googleApiKey: String? = null): String {
-        val actualModel = if (!model.isNullOrEmpty()) model else modelName
+        val requestedModel = if (!model.isNullOrBlank()) model.trim() else modelName
+        val actualModel = if (this == XAI && requestedModel == XAI_LEGACY_MODEL_LABEL) {
+            XAI_THINK_FAST_MODEL
+        } else {
+            requestedModel
+        }
         return when (this) {
             AZURE_OPENAI -> if (isOpenAiGaRealtimeModel(actualModel)) {
                 // Azure GA Realtime endpoint
@@ -37,7 +46,7 @@ enum class RealtimeApiProvider(
                 "wss://$azureEndpoint/openai/realtime?api-version=2025-04-01-preview&deployment=$actualModel"
             }
             OPENAI_DIRECT -> "wss://api.openai.com/v1/realtime?model=$actualModel"
-            XAI -> "wss://api.x.ai/v1/realtime"
+            XAI -> "wss://api.x.ai/v1/realtime?model=$actualModel"
             // v1alpha is required for newer native audio models
             GOOGLE_GEMINI -> "wss://generativelanguage.googleapis.com/ws/google.ai.generativelanguage.v1alpha.GenerativeService.BidiGenerateContent?key=${googleApiKey ?: ""}"
         }
 
@@ -920,19 +920,20 @@ class RealtimeSessionManager @Inject constructor() {
 
             sessionConfig.put("audio", audio)
 
-            // x.ai voice: Use directly if it's an x.ai voice, otherwise map from OpenAI names
-            val xaiVoices = setOf("Ara", "Rex", "Sal", "Eve", "Leo")
-            val xaiVoice = if (voice in xaiVoices) {
-                voice // Already an x.ai voice
+            // x.ai voice IDs are lowercase: eve, ara, rex, sal, leo.
+            val normalizedVoice = voice.lowercase()
+            val xaiVoices = setOf("ara", "rex", "sal", "eve", "leo")
+            val xaiVoice = if (normalizedVoice in xaiVoices) {
+                normalizedVoice
             } else {
                 // Fallback mapping from OpenAI voice names (for backwards compatibility)
-                when (voice.lowercase()) {
-                    "alloy", "ash" -> "Ara"
-                    "echo", "ballad" -> "Rex"
-                    "shimmer", "coral" -> "Eve"
-                    "verse" -> "Sal"
-                    "sage" -> "Leo"
-                    else -> "Ara"
+                when (normalizedVoice) {
+                    "alloy", "ash" -> "ara"
+                    "echo", "ballad" -> "rex"
+                    "shimmer", "coral" -> "eve"
+                    "verse" -> "sal"
+                    "sage" -> "leo"
+                    else -> "ara"
                 }
             }
             sessionConfig.put("voice", xaiVoice)
@@ -941,6 +942,9 @@ class RealtimeSessionManager @Inject constructor() {
             if (settings.isUsingRealtimeAudioInput) {
                 sessionConfig.put("turn_detection", JSONObject().apply {
                     put("type", "server_vad")
+                    put("threshold", settings.vadThreshold.coerceIn(0.1f, 0.9f))
+                    put("prefix_padding_ms", settings.prefixPadding)
+                    put("silence_duration_ms", settings.silenceDuration)
                 })
             }
 
@@ -1147,4 +1151,3 @@ class RealtimeSessionManager @Inject constructor() {
     }
 }
 
-
 
@@ -39,6 +39,15 @@ interface RobotLifecycleBridge {
          * Called when robot focus is lost (Pepper) or simulation paused (Standalone)
          */
         fun onRobotFocusLost()
+
+        /**
+         * Called when QiSDK refuses to grant robot focus (Pepper only).
+         * Surfaces the exact reason reported by the robot service so callers can
+         * display it instead of a generic timeout.
+         *
+         * Default implementation is empty to avoid breaking existing listeners.
+         */
+        fun onRobotFocusRefused(reason: String) {}
     }
 }
 
@@ -14,6 +14,8 @@ import androidx.lifecycle.compose.collectAsStateWithLifecycle
 import ch.fhnw.pepper_realtime.manager.ApiKeyManager
 import ch.fhnw.pepper_realtime.manager.SettingsRepository
 import ch.fhnw.pepper_realtime.network.RealtimeApiProvider
+import ch.fhnw.pepper_realtime.network.XAI_FAST_MODEL
+import ch.fhnw.pepper_realtime.network.XAI_THINK_FAST_MODEL
 import ch.fhnw.pepper_realtime.tools.ToolRegistry
 import ch.fhnw.pepper_realtime.ui.compose.ChatTheme
 import ch.fhnw.pepper_realtime.ui.settings.SettingsViewModel
@@ -61,7 +63,10 @@ fun SettingsScreen(
         "gpt-4o-realtime-preview",
         "gpt-4o-mini-realtime-preview"
     )
-    val xaiModels = listOf("Grok Voice Agent")
+    val xaiModels = listOf(
+        XAI_THINK_FAST_MODEL,
+        XAI_FAST_MODEL
+    )
     // Google Live API requires models/ prefix for BidiGenerateContent
     // Include fallback model for debugging
     val googleModels = listOf(
@@ -88,11 +93,11 @@ fun SettingsScreen(
         "verse" to "Versatile, expressive"
     )
     val xaiVoices = linkedMapOf(
-        "Ara" to "Female, warm, friendly",
-        "Rex" to "Male, confident, clear",
-        "Sal" to "Neutral, smooth, balanced",
-        "Eve" to "Female, energetic, upbeat",
-        "Leo" to "Male, authoritative, strong"
+        "eve" to "Female",
+        "ara" to "Female",
+        "rex" to "Male",
+        "sal" to "Male",
+        "leo" to "Male"
     )
     val googleVoices = linkedMapOf(
         "Zephyr" to "Bright",
@@ -263,12 +268,12 @@ fun SettingsScreen(
             SettingsVoiceDropdown(
                 label = "Voice",
                 voices = voices,
-                selectedVoice = settings.voice,
+                selectedVoice = if (isXaiProvider) settings.voice.lowercase() else settings.voice,
                 onVoiceSelected = { viewModel.setVoice(it) }
             )
 
-            // Speed and Temperature - not supported by Google Live API
-            if (!isGoogleProvider) {
+            // Speed and Temperature - not supported by Google Live API or x.ai Voice Agent
+            if (!isGoogleProvider && !isXaiProvider) {
                 SettingsSlider(
                     label = "Speech Speed",
                     value = settings.speedProgress.toFloat(),
@@ -415,6 +420,33 @@ fun SettingsScreen(
                         checked = settings.googleContextCompression,
                         onCheckedChange = { viewModel.setGoogleContextCompression(it) }
                     )
+                } else if (isXaiProvider) {
+                    // x.ai Voice Agent Settings
+                    SettingsSectionHeader(title = "x.ai Voice Agent Settings")
+
+                    SettingsSlider(
+                        label = "VAD Activation Threshold",
+                        value = settings.vadThreshold.coerceIn(0.1f, 0.9f),
+                        onValueChange = { viewModel.setVadThreshold(it) },
+                        valueRange = 0.1f..0.9f,
+                        valueDisplay = "%.2f".format(settings.vadThreshold.coerceIn(0.1f, 0.9f))
+                    )
+
+                    SettingsSlider(
+                        label = "Prefix Padding",
+                        value = settings.prefixPadding.toFloat(),
+                        onValueChange = { viewModel.setPrefixPadding(it.toInt()) },
+                        valueRange = 0f..1000f,
+                        valueDisplay = "${settings.prefixPadding} ms"
+                    )
+
+                    SettingsSlider(
+                        label = "Silence Duration",
+                        value = settings.silenceDuration.toFloat(),
+                        onValueChange = { viewModel.setSilenceDuration(it.toInt()) },
+                        valueRange = 200f..2000f,
+                        valueDisplay = "${settings.silenceDuration} ms"
+                    )
                 } else {
                     // OpenAI Realtime API Settings
                     SettingsSectionHeader(title = "Voice API Settings")
 
@@ -52,6 +52,7 @@ class RobotLifecycleBridgeImpl : RobotLifecycleBridge, RobotLifecycleCallbacks {
 
     override fun onRobotFocusRefused(reason: String) {
         Log.e(TAG, "🤖 Robot focus refused: $reason")
+        listener?.onRobotFocusRefused(reason)
     }
 }
Original file line number	Diff line number	Diff line change
`@@ -52,6 +52,7 @@ class RobotLifecycleBridgeImpl : RobotLifecycleBridge, RobotLifecycleCallbacks {`
`52`	`52`
`53`	`53`	`override fun onRobotFocusRefused(reason: String) {`
`54`	`54`	`Log.e(TAG, "🤖 Robot focus refused: $reason")`
	`55`	`+ listener?.onRobotFocusRefused(reason)`
`55`	`56`	`}`
`56`	`57`	`}`
`57`	`58`