diff --git a/package/secubox/luci-app-localai/Makefile b/package/secubox/luci-app-localai/Makefile
new file mode 100644
index 00000000..ebf1ab62
--- /dev/null
+++ b/package/secubox/luci-app-localai/Makefile
@@ -0,0 +1,33 @@
+# SPDX-License-Identifier: Apache-2.0
+#
+# Copyright (C) 2025 CyberMind.fr - Gandalf
+#
+# LuCI LocalAI - Self-hosted LLM Management Interface
+#
+
+include $(TOPDIR)/rules.mk
+
+PKG_NAME:=luci-app-localai
+PKG_VERSION:=0.1.0
+PKG_RELEASE:=8
+PKG_ARCH:=all
+
+PKG_LICENSE:=Apache-2.0
+PKG_MAINTAINER:=CyberMind <contact@cybermind.fr>
+
+LUCI_TITLE:=LuCI LocalAI Dashboard
+LUCI_DESCRIPTION:=Modern dashboard for LocalAI LLM management on OpenWrt
+LUCI_DEPENDS:=+luci-base +luci-app-secubox +luci-lib-jsonc +rpcd +rpcd-mod-luci +secubox-app-localai
+
+LUCI_PKGARCH:=all
+
+# File permissions
+PKG_FILE_MODES:=/usr/libexec/rpcd/luci.localai:root:root:755
+
+include $(TOPDIR)/feeds/luci/luci.mk
+
+define Package/$(PKG_NAME)/conffiles
+/etc/config/localai
+endef
+
+# call BuildPackage - OpenWrt buildroot
diff --git a/package/secubox/luci-app-localai/htdocs/luci-static/resources/localai/api.js b/package/secubox/luci-app-localai/htdocs/luci-static/resources/localai/api.js
new file mode 100644
index 00000000..f199e61b
--- /dev/null
+++ b/package/secubox/luci-app-localai/htdocs/luci-static/resources/localai/api.js
@@ -0,0 +1,140 @@
+'use strict';
+'require baseclass';
+'require rpc';
+
+/**
+ * LocalAI Dashboard API
+ * Package: luci-app-localai
+ * RPCD object: luci.localai
+ */
+
+// Version: 0.1.0
+
+var callStatus = rpc.declare({
+	object: 'luci.localai',
+	method: 'status',
+	expect: { }
+});
+
+var callModels = rpc.declare({
+	object: 'luci.localai',
+	method: 'models',
+	expect: { models: [] }
+});
+
+var callConfig = rpc.declare({
+	object: 'luci.localai',
+	method: 'config',
+	expect: { }
+});
+
+var callHealth = rpc.declare({
+	object: 'luci.localai',
+	method: 'health',
+	expect: { healthy: false }
+});
+
+var callMetrics = rpc.declare({
+	object: 'luci.localai',
+	method: 'metrics',
+	expect: { }
+});
+
+var callStart = rpc.declare({
+	object: 'luci.localai',
+	method: 'start',
+	expect: { success: false }
+});
+
+var callStop = rpc.declare({
+	object: 'luci.localai',
+	method: 'stop',
+	expect: { success: false }
+});
+
+var callRestart = rpc.declare({
+	object: 'luci.localai',
+	method: 'restart',
+	expect: { success: false }
+});
+
+var callModelInstall = rpc.declare({
+	object: 'luci.localai',
+	method: 'model_install',
+	params: ['name'],
+	expect: { success: false }
+});
+
+var callModelRemove = rpc.declare({
+	object: 'luci.localai',
+	method: 'model_remove',
+	params: ['name'],
+	expect: { success: false }
+});
+
+var callChat = rpc.declare({
+	object: 'luci.localai',
+	method: 'chat',
+	params: ['model', 'messages'],
+	expect: { response: '' }
+});
+
+var callComplete = rpc.declare({
+	object: 'luci.localai',
+	method: 'complete',
+	params: ['model', 'prompt'],
+	expect: { text: '' }
+});
+
+function formatBytes(bytes) {
+	if (bytes === 0) return '0 B';
+	var k = 1024;
+	var sizes = ['B', 'KB', 'MB', 'GB', 'TB'];
+	var i = Math.floor(Math.log(bytes) / Math.log(k));
+	return parseFloat((bytes / Math.pow(k, i)).toFixed(2)) + ' ' + sizes[i];
+}
+
+function formatUptime(seconds) {
+	if (!seconds) return 'N/A';
+	var days = Math.floor(seconds / 86400);
+	var hours = Math.floor((seconds % 86400) / 3600);
+	var mins = Math.floor((seconds % 3600) / 60);
+
+	if (days > 0) return days + 'd ' + hours + 'h';
+	if (hours > 0) return hours + 'h ' + mins + 'm';
+	return mins + 'm';
+}
+
+return baseclass.extend({
+	getStatus: callStatus,
+	getModels: callModels,
+	getConfig: callConfig,
+	getHealth: callHealth,
+	getMetrics: callMetrics,
+	start: callStart,
+	stop: callStop,
+	restart: callRestart,
+	modelInstall: callModelInstall,
+	modelRemove: callModelRemove,
+	chat: callChat,
+	complete: callComplete,
+	formatBytes: formatBytes,
+	formatUptime: formatUptime,
+
+	// Aggregate function for dashboard
+	getDashboardData: function() {
+		return Promise.all([
+			callStatus(),
+			callModels(),
+			callHealth(),
+			callMetrics()
+		]).then(function(results) {
+			return {
+				status: results[0] || {},
+				models: results[1] || { models: [] },
+				health: results[2] || { healthy: false },
+				metrics: results[3] || {}
+			};
+		});
+	}
+});
diff --git a/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/chat.js b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/chat.js
new file mode 100644
index 00000000..ecf098e4
--- /dev/null
+++ b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/chat.js
@@ -0,0 +1,334 @@
+'use strict';
+'require view';
+'require ui';
+'require rpc';
+
+var callModels = rpc.declare({
+	object: 'luci.localai',
+	method: 'models',
+	expect: { models: [] }
+});
+
+var callChat = rpc.declare({
+	object: 'luci.localai',
+	method: 'chat',
+	params: ['model', 'messages'],
+	expect: { response: '' }
+});
+
+return view.extend({
+	title: _('LocalAI Chat'),
+	messages: [],
+	selectedModel: null,
+
+	load: function() {
+		return callModels();
+	},
+
+	render: function(data) {
+		var self = this;
+		var models = data.models || [];
+
+		var container = E('div', { 'class': 'localai-chat' }, [
+			E('style', {}, this.getCSS()),
+
+			// Header
+			E('div', { 'class': 'chat-header' }, [
+				E('div', { 'class': 'chat-title' }, [
+					E('span', { 'class': 'chat-icon' }, '💬'),
+					_('LocalAI Chat')
+				]),
+				E('div', { 'class': 'chat-model-select' }, [
+					E('label', {}, _('Model:')),
+					E('select', {
+						'id': 'model-select',
+						'change': function(e) { self.selectedModel = e.target.value; }
+					}, models.length > 0 ?
+						models.map(function(m) {
+							var modelId = m.id || m.name;
+							var displayName = m.loaded ? modelId + ' ✓' : modelId;
+							return E('option', { 'value': modelId }, displayName);
+						}) :
+						[E('option', { 'value': '' }, _('No models available'))]
+					)
+				])
+			]),
+
+			// Chat Messages
+			E('div', { 'class': 'chat-messages', 'id': 'chat-messages' }, [
+				E('div', { 'class': 'chat-welcome' }, [
+					E('div', { 'class': 'welcome-icon' }, '🤖'),
+					E('div', { 'class': 'welcome-title' }, _('Welcome to LocalAI Chat')),
+					E('div', { 'class': 'welcome-text' }, _('Start a conversation with your local AI model.'))
+				])
+			]),
+
+			// Input Area
+			E('div', { 'class': 'chat-input-area' }, [
+				E('textarea', {
+					'id': 'chat-input',
+					'class': 'chat-input',
+					'placeholder': _('Type your message...'),
+					'rows': 2,
+					'keydown': function(e) {
+						if (e.key === 'Enter' && !e.shiftKey) {
+							e.preventDefault();
+							self.sendMessage();
+						}
+					}
+				}),
+				E('button', {
+					'class': 'chat-send-btn',
+					'click': function() { self.sendMessage(); }
+				}, [E('span', {}, '➤'), _('Send')])
+			])
+		]);
+
+		// Set initial model
+		if (models.length > 0) {
+			var loadedModel = models.find(function(m) { return m.loaded; });
+			this.selectedModel = loadedModel ? (loadedModel.id || loadedModel.name) : (models[0].id || models[0].name);
+		}
+
+		return container;
+	},
+
+	sendMessage: function() {
+		var self = this;
+		var input = document.getElementById('chat-input');
+		var messagesContainer = document.getElementById('chat-messages');
+		var message = input.value.trim();
+
+		if (!message) return;
+
+		if (!this.selectedModel) {
+			ui.addNotification(null, E('p', _('Please select a model first')), 'error');
+			return;
+		}
+
+		// Clear welcome if present
+		var welcome = messagesContainer.querySelector('.chat-welcome');
+		if (welcome) welcome.remove();
+
+		// Add user message
+		messagesContainer.appendChild(E('div', { 'class': 'chat-message user' }, [
+			E('div', { 'class': 'message-avatar' }, '👤'),
+			E('div', { 'class': 'message-content' }, [
+				E('div', { 'class': 'message-text' }, message)
+			])
+		]));
+
+		// Add loading indicator
+		var loadingMsg = E('div', { 'class': 'chat-message assistant loading', 'id': 'loading-msg' }, [
+			E('div', { 'class': 'message-avatar' }, '🤖'),
+			E('div', { 'class': 'message-content' }, [
+				E('div', { 'class': 'message-loading' }, [
+					E('span', {}), E('span', {}), E('span', {})
+				])
+			])
+		]);
+		messagesContainer.appendChild(loadingMsg);
+		messagesContainer.scrollTop = messagesContainer.scrollHeight;
+
+		// Clear input
+		input.value = '';
+
+		// Build messages array
+		this.messages.push({ role: 'user', content: message });
+
+		// Send to API
+		callChat(this.selectedModel, JSON.stringify(this.messages))
+			.then(function(result) {
+				var loading = document.getElementById('loading-msg');
+				if (loading) loading.remove();
+
+				var response = result.response || result.error || _('No response');
+				self.messages.push({ role: 'assistant', content: response });
+
+				messagesContainer.appendChild(E('div', { 'class': 'chat-message assistant' }, [
+					E('div', { 'class': 'message-avatar' }, '🤖'),
+					E('div', { 'class': 'message-content' }, [
+						E('div', { 'class': 'message-text' }, response)
+					])
+				]));
+				messagesContainer.scrollTop = messagesContainer.scrollHeight;
+			})
+			.catch(function(err) {
+				var loading = document.getElementById('loading-msg');
+				if (loading) loading.remove();
+
+				messagesContainer.appendChild(E('div', { 'class': 'chat-message assistant error' }, [
+					E('div', { 'class': 'message-avatar' }, '⚠️'),
+					E('div', { 'class': 'message-content' }, [
+						E('div', { 'class': 'message-text' }, _('Error: ') + err.message)
+					])
+				]));
+				messagesContainer.scrollTop = messagesContainer.scrollHeight;
+			});
+	},
+
+	getCSS: function() {
+		return `
+			.localai-chat {
+				font-family: 'Inter', -apple-system, sans-serif;
+				background: #030712;
+				color: #f8fafc;
+				min-height: calc(100vh - 100px);
+				display: flex;
+				flex-direction: column;
+				padding: 16px;
+			}
+			.chat-header {
+				display: flex;
+				align-items: center;
+				justify-content: space-between;
+				padding: 16px 20px;
+				background: #0f172a;
+				border: 1px solid #334155;
+				border-radius: 12px;
+				margin-bottom: 16px;
+			}
+			.chat-title {
+				display: flex;
+				align-items: center;
+				gap: 12px;
+				font-size: 18px;
+				font-weight: 600;
+			}
+			.chat-icon { font-size: 24px; }
+			.chat-model-select {
+				display: flex;
+				align-items: center;
+				gap: 10px;
+			}
+			.chat-model-select select {
+				padding: 8px 12px;
+				background: #1e293b;
+				border: 1px solid #334155;
+				border-radius: 8px;
+				color: #f8fafc;
+				font-size: 13px;
+			}
+			.chat-messages {
+				flex: 1;
+				overflow-y: auto;
+				padding: 20px;
+				background: #0f172a;
+				border: 1px solid #334155;
+				border-radius: 12px;
+				margin-bottom: 16px;
+				display: flex;
+				flex-direction: column;
+				gap: 16px;
+				min-height: 400px;
+			}
+			.chat-welcome {
+				text-align: center;
+				padding: 60px 20px;
+				color: #64748b;
+			}
+			.welcome-icon { font-size: 64px; margin-bottom: 20px; }
+			.welcome-title { font-size: 20px; font-weight: 600; margin-bottom: 10px; color: #94a3b8; }
+			.welcome-text { font-size: 14px; }
+			.chat-message {
+				display: flex;
+				gap: 12px;
+				max-width: 80%;
+			}
+			.chat-message.user {
+				align-self: flex-end;
+				flex-direction: row-reverse;
+			}
+			.message-avatar {
+				width: 36px;
+				height: 36px;
+				border-radius: 10px;
+				background: #1e293b;
+				display: flex;
+				align-items: center;
+				justify-content: center;
+				font-size: 18px;
+				flex-shrink: 0;
+			}
+			.chat-message.user .message-avatar {
+				background: linear-gradient(135deg, #06b6d4, #0ea5e9);
+			}
+			.chat-message.assistant .message-avatar {
+				background: linear-gradient(135deg, #a855f7, #6366f1);
+			}
+			.message-content {
+				padding: 12px 16px;
+				border-radius: 12px;
+				background: #1e293b;
+			}
+			.chat-message.user .message-content {
+				background: linear-gradient(135deg, #06b6d4, #0ea5e9);
+			}
+			.chat-message.error .message-content {
+				background: rgba(239, 68, 68, 0.2);
+				border: 1px solid rgba(239, 68, 68, 0.3);
+			}
+			.message-text {
+				font-size: 14px;
+				line-height: 1.5;
+				white-space: pre-wrap;
+			}
+			.message-loading {
+				display: flex;
+				gap: 4px;
+			}
+			.message-loading span {
+				width: 8px;
+				height: 8px;
+				background: #a855f7;
+				border-radius: 50%;
+				animation: bounce 1.4s infinite ease-in-out;
+			}
+			.message-loading span:nth-child(1) { animation-delay: -0.32s; }
+			.message-loading span:nth-child(2) { animation-delay: -0.16s; }
+			@keyframes bounce {
+				0%, 80%, 100% { transform: scale(0); }
+				40% { transform: scale(1); }
+			}
+			.chat-input-area {
+				display: flex;
+				gap: 12px;
+				padding: 16px;
+				background: #0f172a;
+				border: 1px solid #334155;
+				border-radius: 12px;
+			}
+			.chat-input {
+				flex: 1;
+				padding: 12px 16px;
+				background: #1e293b;
+				border: 1px solid #334155;
+				border-radius: 10px;
+				color: #f8fafc;
+				font-size: 14px;
+				resize: none;
+				font-family: inherit;
+			}
+			.chat-input:focus {
+				outline: none;
+				border-color: #a855f7;
+			}
+			.chat-send-btn {
+				display: flex;
+				align-items: center;
+				gap: 8px;
+				padding: 12px 24px;
+				background: linear-gradient(135deg, #a855f7, #6366f1);
+				border: none;
+				border-radius: 10px;
+				color: white;
+				font-size: 14px;
+				font-weight: 600;
+				cursor: pointer;
+			}
+			.chat-send-btn:hover {
+				box-shadow: 0 0 20px rgba(168, 85, 247, 0.4);
+			}
+		`;
+	}
+});
diff --git a/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/dashboard.js b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/dashboard.js
new file mode 100644
index 00000000..ab83fe29
--- /dev/null
+++ b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/dashboard.js
@@ -0,0 +1,602 @@
+'use strict';
+'require view';
+'require ui';
+'require rpc';
+
+var callStatus = rpc.declare({
+	object: 'luci.localai',
+	method: 'status',
+	expect: { }
+});
+
+var callModels = rpc.declare({
+	object: 'luci.localai',
+	method: 'models',
+	expect: { models: [] }
+});
+
+var callHealth = rpc.declare({
+	object: 'luci.localai',
+	method: 'health',
+	expect: { healthy: false }
+});
+
+var callMetrics = rpc.declare({
+	object: 'luci.localai',
+	method: 'metrics',
+	expect: { }
+});
+
+var callStart = rpc.declare({
+	object: 'luci.localai',
+	method: 'start',
+	expect: { success: false }
+});
+
+var callStop = rpc.declare({
+	object: 'luci.localai',
+	method: 'stop',
+	expect: { success: false }
+});
+
+var callRestart = rpc.declare({
+	object: 'luci.localai',
+	method: 'restart',
+	expect: { success: false }
+});
+
+function formatBytes(bytes) {
+	if (!bytes || bytes === 0) return '0 B';
+	var k = 1024;
+	var sizes = ['B', 'KB', 'MB', 'GB', 'TB'];
+	var i = Math.floor(Math.log(bytes) / Math.log(k));
+	return parseFloat((bytes / Math.pow(k, i)).toFixed(2)) + ' ' + sizes[i];
+}
+
+function formatUptime(seconds) {
+	if (!seconds) return 'N/A';
+	var days = Math.floor(seconds / 86400);
+	var hours = Math.floor((seconds % 86400) / 3600);
+	var mins = Math.floor((seconds % 3600) / 60);
+	if (days > 0) return days + 'd ' + hours + 'h';
+	if (hours > 0) return hours + 'h ' + mins + 'm';
+	return mins + 'm';
+}
+
+return view.extend({
+	title: _('LocalAI Dashboard'),
+	refreshInterval: 5000,
+	data: null,
+
+	load: function() {
+		return Promise.all([
+			callStatus(),
+			callModels(),
+			callHealth(),
+			callMetrics()
+		]).then(function(results) {
+			return {
+				status: results[0] || {},
+				models: results[1] || { models: [] },
+				health: results[2] || { healthy: false },
+				metrics: results[3] || {}
+			};
+		});
+	},
+
+	render: function(data) {
+		var self = this;
+		this.data = data;
+
+		var container = E('div', { 'class': 'localai-dashboard' }, [
+			// Header
+			E('div', { 'class': 'lai-header' }, [
+				E('div', { 'class': 'lai-logo' }, [
+					E('div', { 'class': 'lai-logo-icon' }, '🤖'),
+					E('div', { 'class': 'lai-logo-text' }, [
+						E('span', {}, 'Local'),
+						'AI'
+					])
+				]),
+				E('div', { 'class': 'lai-header-info' }, [
+					E('div', {
+						'class': 'lai-status-badge ' + (data.status.running ? '' : 'offline'),
+						'id': 'lai-status-badge'
+					}, [
+						E('span', { 'class': 'lai-status-dot' }),
+						data.status.running ? _('Running') : _('Stopped')
+					])
+				])
+			]),
+
+			// Quick Stats
+			E('div', { 'class': 'lai-quick-stats' }, [
+				E('div', { 'class': 'lai-quick-stat', 'style': '--stat-gradient: linear-gradient(135deg, #a855f7, #6366f1)' }, [
+					E('div', { 'class': 'lai-quick-stat-header' }, [
+						E('span', { 'class': 'lai-quick-stat-icon' }, '📊'),
+						E('span', { 'class': 'lai-quick-stat-label' }, _('Models'))
+					]),
+					E('div', { 'class': 'lai-quick-stat-value', 'id': 'models-count' },
+						(data.models.models || []).length.toString()
+					),
+					E('div', { 'class': 'lai-quick-stat-sub' }, _('Installed'))
+				]),
+
+				E('div', { 'class': 'lai-quick-stat', 'style': '--stat-gradient: linear-gradient(135deg, #10b981, #059669)' }, [
+					E('div', { 'class': 'lai-quick-stat-header' }, [
+						E('span', { 'class': 'lai-quick-stat-icon' }, '💾'),
+						E('span', { 'class': 'lai-quick-stat-label' }, _('Memory'))
+					]),
+					E('div', { 'class': 'lai-quick-stat-value', 'id': 'memory-used' },
+						formatBytes(data.metrics.memory_used || 0)
+					),
+					E('div', { 'class': 'lai-quick-stat-sub' }, _('Used'))
+				]),
+
+				E('div', { 'class': 'lai-quick-stat', 'style': '--stat-gradient: linear-gradient(135deg, #06b6d4, #0ea5e9)' }, [
+					E('div', { 'class': 'lai-quick-stat-header' }, [
+						E('span', { 'class': 'lai-quick-stat-icon' }, '⏱️'),
+						E('span', { 'class': 'lai-quick-stat-label' }, _('Uptime'))
+					]),
+					E('div', { 'class': 'lai-quick-stat-value', 'id': 'uptime' },
+						data.status.running ? formatUptime(data.status.uptime) : '--'
+					),
+					E('div', { 'class': 'lai-quick-stat-sub' }, _('Running'))
+				]),
+
+				E('div', { 'class': 'lai-quick-stat', 'style': '--stat-gradient: linear-gradient(135deg, #f59e0b, #d97706)' }, [
+					E('div', { 'class': 'lai-quick-stat-header' }, [
+						E('span', { 'class': 'lai-quick-stat-icon' }, '🔌'),
+						E('span', { 'class': 'lai-quick-stat-label' }, _('API Port'))
+					]),
+					E('div', { 'class': 'lai-quick-stat-value' }, data.status.api_port || '8080'),
+					E('div', { 'class': 'lai-quick-stat-sub' }, _('Endpoint'))
+				])
+			]),
+
+			// Main Cards Grid
+			E('div', { 'class': 'lai-cards-grid' }, [
+				// Service Control Card
+				E('div', { 'class': 'lai-card' }, [
+					E('div', { 'class': 'lai-card-header' }, [
+						E('div', { 'class': 'lai-card-title' }, [
+							E('span', { 'class': 'lai-card-title-icon' }, '⚙️'),
+							_('Service Control')
+						]),
+						E('div', {
+							'class': 'lai-card-badge ' + (data.status.running ? 'running' : 'stopped')
+						}, data.status.running ? _('Active') : _('Inactive'))
+					]),
+					E('div', { 'class': 'lai-card-body' }, [
+						E('div', { 'class': 'lai-service-info' }, [
+							E('div', { 'class': 'lai-service-row' }, [
+								E('span', { 'class': 'lai-service-label' }, _('Status')),
+								E('span', {
+									'class': 'lai-service-value ' + (data.status.running ? 'running' : 'stopped'),
+									'id': 'service-status'
+								}, data.status.running ? _('Running') : _('Stopped'))
+							]),
+							E('div', { 'class': 'lai-service-row' }, [
+								E('span', { 'class': 'lai-service-label' }, _('Memory Limit')),
+								E('span', { 'class': 'lai-service-value' }, data.status.memory_limit || '2G')
+							]),
+							E('div', { 'class': 'lai-service-row' }, [
+								E('span', { 'class': 'lai-service-label' }, _('Threads')),
+								E('span', { 'class': 'lai-service-value' }, data.status.threads || '4')
+							]),
+							E('div', { 'class': 'lai-service-row' }, [
+								E('span', { 'class': 'lai-service-label' }, _('Context Size')),
+								E('span', { 'class': 'lai-service-value' }, data.status.context_size || '2048')
+							])
+						]),
+						E('div', { 'class': 'lai-service-controls' }, [
+							E('button', {
+								'class': 'lai-btn lai-btn-success' + (data.status.running ? ' disabled' : ''),
+								'click': function() { self.handleServiceAction('start'); },
+								'disabled': data.status.running
+							}, [E('span', {}, '▶'), _('Start')]),
+							E('button', {
+								'class': 'lai-btn lai-btn-danger' + (!data.status.running ? ' disabled' : ''),
+								'click': function() { self.handleServiceAction('stop'); },
+								'disabled': !data.status.running
+							}, [E('span', {}, '⏹'), _('Stop')]),
+							E('button', {
+								'class': 'lai-btn lai-btn-warning',
+								'click': function() { self.handleServiceAction('restart'); }
+							}, [E('span', {}, '🔄'), _('Restart')])
+						])
+					])
+				]),
+
+				// Models Card
+				E('div', { 'class': 'lai-card' }, [
+					E('div', { 'class': 'lai-card-header' }, [
+						E('div', { 'class': 'lai-card-title' }, [
+							E('span', { 'class': 'lai-card-title-icon' }, '🧠'),
+							_('Installed Models')
+						]),
+						E('div', { 'class': 'lai-card-badge' },
+							(data.models.models || []).length + ' ' + _('models')
+						)
+					]),
+					E('div', { 'class': 'lai-card-body' }, [
+						this.renderModelsList(data.models.models || [])
+					])
+				])
+			]),
+
+			// API Info Card
+			E('div', { 'class': 'lai-card', 'style': 'margin-top: 20px' }, [
+				E('div', { 'class': 'lai-card-header' }, [
+					E('div', { 'class': 'lai-card-title' }, [
+						E('span', { 'class': 'lai-card-title-icon' }, '🔗'),
+						_('API Endpoints')
+					])
+				]),
+				E('div', { 'class': 'lai-card-body' }, [
+					E('div', { 'class': 'lai-api-info' }, [
+						E('div', { 'class': 'lai-api-endpoint' }, [
+							E('code', {}, 'http://' + window.location.hostname + ':' + (data.status.api_port || '8080') + '/v1/chat/completions'),
+							E('span', { 'class': 'lai-api-method' }, 'POST'),
+							E('span', { 'class': 'lai-api-desc' }, _('Chat completion'))
+						]),
+						E('div', { 'class': 'lai-api-endpoint' }, [
+							E('code', {}, 'http://' + window.location.hostname + ':' + (data.status.api_port || '8080') + '/v1/models'),
+							E('span', { 'class': 'lai-api-method get' }, 'GET'),
+							E('span', { 'class': 'lai-api-desc' }, _('List models'))
+						])
+					])
+				])
+			])
+		]);
+
+		// Include CSS
+		var style = E('style', {}, this.getCSS());
+		container.insertBefore(style, container.firstChild);
+
+		return container;
+	},
+
+	renderModelsList: function(models) {
+		if (!models || models.length === 0) {
+			return E('div', { 'class': 'lai-empty' }, [
+				E('div', { 'class': 'lai-empty-icon' }, '📦'),
+				E('div', { 'class': 'lai-empty-text' }, _('No models installed')),
+				E('div', { 'class': 'lai-empty-hint' }, [
+					_('Install a model with: '),
+					E('code', {}, 'localaictl model-install tinyllama')
+				])
+			]);
+		}
+
+		return E('div', { 'class': 'lai-models-list' },
+			models.map(function(model) {
+				var displayName = model.id || model.name;
+				return E('div', { 'class': 'lai-model-item' + (model.loaded ? ' loaded' : '') }, [
+					E('div', { 'class': 'lai-model-icon' }, model.loaded ? '✅' : '🤖'),
+					E('div', { 'class': 'lai-model-info' }, [
+						E('div', { 'class': 'lai-model-name' }, displayName),
+						E('div', { 'class': 'lai-model-meta' }, [
+							model.size > 0 ? E('span', { 'class': 'lai-model-size' }, formatBytes(model.size)) : null,
+							E('span', { 'class': 'lai-model-type' }, model.loaded ? _('Active') : model.type)
+						].filter(Boolean))
+					])
+				]);
+			})
+		);
+	},
+
+	handleServiceAction: function(action) {
+		var self = this;
+
+		ui.showModal(_('Service Control'), [
+			E('p', {}, _('Processing...')),
+			E('div', { 'class': 'spinning' })
+		]);
+
+		var actionFn;
+		switch(action) {
+			case 'start': actionFn = callStart(); break;
+			case 'stop': actionFn = callStop(); break;
+			case 'restart': actionFn = callRestart(); break;
+		}
+
+		actionFn.then(function(result) {
+			ui.hideModal();
+			if (result.success) {
+				ui.addNotification(null, E('p', _('Service ' + action + ' successful')), 'success');
+				window.location.reload();
+			} else {
+				ui.addNotification(null, E('p', result.error || _('Operation failed')), 'error');
+			}
+		}).catch(function(err) {
+			ui.hideModal();
+			ui.addNotification(null, E('p', err.message), 'error');
+		});
+	},
+
+	getCSS: function() {
+		return `
+			.localai-dashboard {
+				font-family: 'Inter', -apple-system, BlinkMacSystemFont, sans-serif;
+				background: #030712;
+				color: #f8fafc;
+				min-height: 100vh;
+				padding: 16px;
+			}
+			.lai-header {
+				display: flex;
+				align-items: center;
+				justify-content: space-between;
+				padding: 12px 0 20px;
+				border-bottom: 1px solid #334155;
+				margin-bottom: 20px;
+			}
+			.lai-logo {
+				display: flex;
+				align-items: center;
+				gap: 14px;
+			}
+			.lai-logo-icon {
+				width: 46px;
+				height: 46px;
+				background: linear-gradient(135deg, #a855f7, #6366f1);
+				border-radius: 12px;
+				display: flex;
+				align-items: center;
+				justify-content: center;
+				font-size: 24px;
+			}
+			.lai-logo-text {
+				font-size: 24px;
+				font-weight: 700;
+			}
+			.lai-logo-text span {
+				background: linear-gradient(135deg, #a855f7, #6366f1);
+				-webkit-background-clip: text;
+				-webkit-text-fill-color: transparent;
+			}
+			.lai-status-badge {
+				display: flex;
+				align-items: center;
+				gap: 8px;
+				padding: 8px 16px;
+				border-radius: 24px;
+				background: rgba(16, 185, 129, 0.15);
+				color: #10b981;
+				border: 1px solid rgba(16, 185, 129, 0.3);
+				font-weight: 600;
+			}
+			.lai-status-badge.offline {
+				background: rgba(239, 68, 68, 0.15);
+				color: #ef4444;
+				border-color: rgba(239, 68, 68, 0.3);
+			}
+			.lai-status-dot {
+				width: 10px;
+				height: 10px;
+				background: currentColor;
+				border-radius: 50%;
+			}
+			.lai-quick-stats {
+				display: grid;
+				grid-template-columns: repeat(auto-fit, minmax(180px, 1fr));
+				gap: 14px;
+				margin-bottom: 24px;
+			}
+			.lai-quick-stat {
+				background: #0f172a;
+				border: 1px solid #334155;
+				border-radius: 12px;
+				padding: 20px;
+				position: relative;
+				overflow: hidden;
+			}
+			.lai-quick-stat::before {
+				content: '';
+				position: absolute;
+				top: 0;
+				left: 0;
+				right: 0;
+				height: 3px;
+				background: var(--stat-gradient);
+			}
+			.lai-quick-stat-header {
+				display: flex;
+				align-items: center;
+				gap: 10px;
+				margin-bottom: 12px;
+			}
+			.lai-quick-stat-icon { font-size: 22px; }
+			.lai-quick-stat-label {
+				font-size: 11px;
+				text-transform: uppercase;
+				color: #64748b;
+			}
+			.lai-quick-stat-value {
+				font-size: 32px;
+				font-weight: 700;
+				background: var(--stat-gradient);
+				-webkit-background-clip: text;
+				-webkit-text-fill-color: transparent;
+			}
+			.lai-quick-stat-sub {
+				font-size: 11px;
+				color: #64748b;
+				margin-top: 6px;
+			}
+			.lai-cards-grid {
+				display: grid;
+				grid-template-columns: repeat(auto-fit, minmax(400px, 1fr));
+				gap: 20px;
+			}
+			.lai-card {
+				background: #0f172a;
+				border: 1px solid #334155;
+				border-radius: 12px;
+				overflow: hidden;
+			}
+			.lai-card-header {
+				display: flex;
+				align-items: center;
+				justify-content: space-between;
+				padding: 16px 20px;
+				border-bottom: 1px solid #334155;
+				background: rgba(0, 0, 0, 0.3);
+			}
+			.lai-card-title {
+				display: flex;
+				align-items: center;
+				gap: 12px;
+				font-size: 15px;
+				font-weight: 600;
+			}
+			.lai-card-title-icon { font-size: 20px; }
+			.lai-card-badge {
+				font-size: 12px;
+				padding: 5px 12px;
+				border-radius: 16px;
+				background: linear-gradient(135deg, #a855f7, #6366f1);
+				color: white;
+			}
+			.lai-card-badge.running { background: linear-gradient(135deg, #10b981, #059669); }
+			.lai-card-badge.stopped { background: rgba(100, 116, 139, 0.3); color: #94a3b8; }
+			.lai-card-body { padding: 20px; }
+			.lai-service-info {
+				display: flex;
+				flex-direction: column;
+				gap: 12px;
+				margin-bottom: 20px;
+			}
+			.lai-service-row {
+				display: flex;
+				justify-content: space-between;
+				padding: 8px 12px;
+				background: #030712;
+				border-radius: 8px;
+			}
+			.lai-service-label { color: #94a3b8; font-size: 13px; }
+			.lai-service-value { font-size: 13px; }
+			.lai-service-value.running { color: #10b981; }
+			.lai-service-value.stopped { color: #ef4444; }
+			.lai-service-controls {
+				display: flex;
+				gap: 10px;
+			}
+			.lai-btn {
+				display: inline-flex;
+				align-items: center;
+				gap: 6px;
+				padding: 10px 16px;
+				border: none;
+				border-radius: 8px;
+				font-size: 13px;
+				font-weight: 500;
+				cursor: pointer;
+			}
+			.lai-btn-success {
+				background: linear-gradient(135deg, #10b981, #059669);
+				color: white;
+			}
+			.lai-btn-danger {
+				background: linear-gradient(135deg, #ef4444, #dc2626);
+				color: white;
+			}
+			.lai-btn-warning {
+				background: linear-gradient(135deg, #f59e0b, #d97706);
+				color: white;
+			}
+			.lai-btn.disabled {
+				opacity: 0.5;
+				cursor: not-allowed;
+			}
+			.lai-models-list {
+				display: flex;
+				flex-direction: column;
+				gap: 12px;
+			}
+			.lai-model-item {
+				display: flex;
+				align-items: center;
+				gap: 14px;
+				padding: 14px;
+				background: #1e293b;
+				border-radius: 10px;
+			}
+			.lai-model-item.loaded {
+				border: 1px solid rgba(16, 185, 129, 0.3);
+				background: rgba(16, 185, 129, 0.05);
+			}
+			.lai-model-icon {
+				width: 44px;
+				height: 44px;
+				background: linear-gradient(135deg, #a855f7, #6366f1);
+				border-radius: 10px;
+				display: flex;
+				align-items: center;
+				justify-content: center;
+				font-size: 20px;
+			}
+			.lai-model-name {
+				font-weight: 600;
+				margin-bottom: 4px;
+			}
+			.lai-model-meta {
+				display: flex;
+				gap: 12px;
+				font-size: 12px;
+				color: #94a3b8;
+			}
+			.lai-model-type {
+				padding: 2px 8px;
+				background: #334155;
+				border-radius: 4px;
+			}
+			.lai-empty {
+				text-align: center;
+				padding: 40px 20px;
+				color: #64748b;
+			}
+			.lai-empty-icon { font-size: 48px; margin-bottom: 12px; }
+			.lai-empty-text { font-size: 16px; margin-bottom: 8px; }
+			.lai-empty-hint { font-size: 13px; }
+			.lai-empty-hint code {
+				background: #1e293b;
+				padding: 4px 8px;
+				border-radius: 4px;
+			}
+			.lai-api-info {
+				display: flex;
+				flex-direction: column;
+				gap: 10px;
+			}
+			.lai-api-endpoint {
+				display: flex;
+				align-items: center;
+				gap: 12px;
+				padding: 12px;
+				background: #030712;
+				border-radius: 8px;
+			}
+			.lai-api-endpoint code {
+				font-size: 12px;
+				color: #06b6d4;
+				flex: 1;
+			}
+			.lai-api-method {
+				padding: 4px 8px;
+				background: #f59e0b;
+				color: #030712;
+				border-radius: 4px;
+				font-size: 10px;
+				font-weight: 700;
+			}
+			.lai-api-method.get { background: #10b981; }
+			.lai-api-desc {
+				font-size: 12px;
+				color: #94a3b8;
+				min-width: 120px;
+			}
+		`;
+	}
+});
diff --git a/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/models.js b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/models.js
new file mode 100644
index 00000000..fcb960b4
--- /dev/null
+++ b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/models.js
@@ -0,0 +1,255 @@
+'use strict';
+'require view';
+'require ui';
+'require rpc';
+
+var callModels = rpc.declare({
+	object: 'luci.localai',
+	method: 'models',
+	expect: { models: [] }
+});
+
+var callModelInstall = rpc.declare({
+	object: 'luci.localai',
+	method: 'model_install',
+	params: ['name'],
+	expect: { success: false }
+});
+
+var callModelRemove = rpc.declare({
+	object: 'luci.localai',
+	method: 'model_remove',
+	params: ['name'],
+	expect: { success: false }
+});
+
+function formatBytes(bytes) {
+	if (!bytes || bytes === 0) return '0 B';
+	var k = 1024;
+	var sizes = ['B', 'KB', 'MB', 'GB', 'TB'];
+	var i = Math.floor(Math.log(bytes) / Math.log(k));
+	return parseFloat((bytes / Math.pow(k, i)).toFixed(2)) + ' ' + sizes[i];
+}
+
+return view.extend({
+	title: _('LocalAI Models'),
+
+	load: function() {
+		return callModels();
+	},
+
+	render: function(data) {
+		var self = this;
+		var models = data.models || [];
+
+		var presets = [
+			{ name: 'tinyllama', desc: 'TinyLlama 1.1B - Ultra-lightweight', size: '669 MB' },
+			{ name: 'phi2', desc: 'Microsoft Phi-2 - Compact and efficient', size: '1.6 GB' },
+			{ name: 'mistral', desc: 'Mistral 7B Instruct - High quality assistant', size: '4.1 GB' },
+			{ name: 'gte-small', desc: 'GTE Small - Fast embeddings', size: '67 MB' }
+		];
+
+		return E('div', { 'class': 'localai-models' }, [
+			E('style', {}, this.getCSS()),
+
+			E('div', { 'class': 'models-header' }, [
+				E('h2', {}, [E('span', {}, '🧠 '), _('Model Management')]),
+				E('p', {}, _('Install and manage AI models for LocalAI'))
+			]),
+
+			// Installed Models
+			E('div', { 'class': 'models-section' }, [
+				E('h3', {}, _('Installed Models')),
+				models.length > 0 ?
+					E('div', { 'class': 'models-grid' },
+						models.map(function(m) {
+							var displayId = m.id || m.name;
+							return E('div', { 'class': 'model-card installed' + (m.loaded ? ' active' : '') }, [
+								E('div', { 'class': 'model-card-icon' }, m.loaded ? '✅' : '🤖'),
+								E('div', { 'class': 'model-card-info' }, [
+									E('div', { 'class': 'model-card-name' }, displayId),
+									E('div', { 'class': 'model-card-meta' }, [
+										m.size > 0 ? E('span', {}, formatBytes(m.size)) : null,
+										E('span', {}, m.loaded ? _('Loaded') : m.type)
+									].filter(Boolean))
+								]),
+								E('button', {
+									'class': 'model-btn danger',
+									'click': function() { self.removeModel(m.name || displayId); }
+								}, _('Remove'))
+							]);
+						})
+					) :
+					E('div', { 'class': 'empty-state' }, [
+						E('span', {}, '📦'),
+						E('p', {}, _('No models installed yet'))
+					])
+			]),
+
+			// Available Presets
+			E('div', { 'class': 'models-section' }, [
+				E('h3', {}, _('Available Presets')),
+				E('div', { 'class': 'models-grid' },
+					presets.map(function(p) {
+						var isInstalled = models.some(function(m) {
+							var mId = (m.id || '').toLowerCase();
+							var mName = (m.name || '').toLowerCase();
+							return mId.includes(p.name) || mName.includes(p.name);
+						});
+
+						return E('div', { 'class': 'model-card preset' + (isInstalled ? ' installed' : '') }, [
+							E('div', { 'class': 'model-card-icon' }, isInstalled ? '✅' : '📥'),
+							E('div', { 'class': 'model-card-info' }, [
+								E('div', { 'class': 'model-card-name' }, p.name),
+								E('div', { 'class': 'model-card-desc' }, p.desc),
+								E('div', { 'class': 'model-card-size' }, p.size)
+							]),
+							!isInstalled ?
+								E('button', {
+									'class': 'model-btn install',
+									'click': function() { self.installModel(p.name); }
+								}, _('Install')) :
+								E('span', { 'class': 'model-installed-badge' }, _('Installed'))
+						]);
+					})
+				)
+			])
+		]);
+	},
+
+	installModel: function(name) {
+		ui.showModal(_('Installing Model'), [
+			E('p', {}, _('Downloading and installing model: ') + name),
+			E('p', { 'class': 'note' }, _('This may take several minutes...')),
+			E('div', { 'class': 'spinning' })
+		]);
+
+		callModelInstall(name).then(function(result) {
+			ui.hideModal();
+			if (result.success) {
+				ui.addNotification(null, E('p', _('Model installed successfully')), 'success');
+				window.location.reload();
+			} else {
+				ui.addNotification(null, E('p', result.error || _('Installation failed')), 'error');
+			}
+		}).catch(function(err) {
+			ui.hideModal();
+			ui.addNotification(null, E('p', err.message), 'error');
+		});
+	},
+
+	removeModel: function(name) {
+		var self = this;
+
+		ui.showModal(_('Remove Model'), [
+			E('p', {}, _('Remove model: ') + name + '?'),
+			E('div', { 'class': 'right' }, [
+				E('button', { 'class': 'btn', 'click': ui.hideModal }, _('Cancel')),
+				E('button', {
+					'class': 'btn danger',
+					'click': function() {
+						callModelRemove(name).then(function(result) {
+							ui.hideModal();
+							if (result.success) {
+								ui.addNotification(null, E('p', _('Model removed')), 'success');
+								window.location.reload();
+							} else {
+								ui.addNotification(null, E('p', result.error || _('Removal failed')), 'error');
+							}
+						});
+					}
+				}, _('Remove'))
+			])
+		]);
+	},
+
+	getCSS: function() {
+		return `
+			.localai-models {
+				font-family: 'Inter', -apple-system, sans-serif;
+				background: #030712;
+				color: #f8fafc;
+				min-height: 100vh;
+				padding: 20px;
+			}
+			.models-header { margin-bottom: 30px; }
+			.models-header h2 { font-size: 24px; margin-bottom: 8px; }
+			.models-header p { color: #94a3b8; }
+			.models-section {
+				background: #0f172a;
+				border: 1px solid #334155;
+				border-radius: 12px;
+				padding: 20px;
+				margin-bottom: 20px;
+			}
+			.models-section h3 {
+				margin: 0 0 16px 0;
+				font-size: 16px;
+				color: #a855f7;
+			}
+			.models-grid {
+				display: grid;
+				grid-template-columns: repeat(auto-fill, minmax(300px, 1fr));
+				gap: 16px;
+			}
+			.model-card {
+				display: flex;
+				align-items: center;
+				gap: 14px;
+				padding: 16px;
+				background: #1e293b;
+				border: 1px solid #334155;
+				border-radius: 10px;
+			}
+			.model-card.active {
+				border-color: rgba(16, 185, 129, 0.4);
+				background: rgba(16, 185, 129, 0.08);
+			}
+			.model-card-icon {
+				width: 48px;
+				height: 48px;
+				background: linear-gradient(135deg, #a855f7, #6366f1);
+				border-radius: 10px;
+				display: flex;
+				align-items: center;
+				justify-content: center;
+				font-size: 24px;
+			}
+			.model-card-info { flex: 1; }
+			.model-card-name { font-weight: 600; margin-bottom: 4px; }
+			.model-card-desc { font-size: 12px; color: #94a3b8; }
+			.model-card-size { font-size: 11px; color: #64748b; margin-top: 4px; }
+			.model-card-meta {
+				display: flex;
+				gap: 10px;
+				font-size: 12px;
+				color: #94a3b8;
+			}
+			.model-btn {
+				padding: 8px 16px;
+				border: none;
+				border-radius: 6px;
+				font-size: 13px;
+				cursor: pointer;
+			}
+			.model-btn.install {
+				background: linear-gradient(135deg, #10b981, #059669);
+				color: white;
+			}
+			.model-btn.danger {
+				background: linear-gradient(135deg, #ef4444, #dc2626);
+				color: white;
+			}
+			.model-installed-badge {
+				font-size: 12px;
+				color: #10b981;
+			}
+			.empty-state {
+				text-align: center;
+				padding: 40px;
+				color: #64748b;
+			}
+			.empty-state span { font-size: 48px; }
+		`;
+	}
+});
diff --git a/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/settings.js b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/settings.js
new file mode 100644
index 00000000..a5a7487e
--- /dev/null
+++ b/package/secubox/luci-app-localai/htdocs/luci-static/resources/view/localai/settings.js
@@ -0,0 +1,91 @@
+'use strict';
+'require view';
+'require form';
+'require uci';
+
+return view.extend({
+	load: function() {
+		return uci.load('localai');
+	},
+
+	render: function() {
+		var m, s, o;
+
+		m = new form.Map('localai', _('LocalAI Configuration'),
+			_('Configure LocalAI service settings. Changes require a service restart.'));
+
+		// Main settings section with tabs
+		s = m.section(form.TypedSection, 'main', _('Service Settings'));
+		s.anonymous = true;
+		s.addremove = false;
+		s.tab('general', _('General'));
+		s.tab('paths', _('Storage'));
+
+		// General tab options
+		o = s.taboption('general', form.Flag, 'enabled', _('Enable Service'));
+		o.default = '0';
+		o.rmempty = false;
+
+		o = s.taboption('general', form.Value, 'api_port', _('API Port'));
+		o.datatype = 'port';
+		o.default = '8080';
+		o.placeholder = '8080';
+
+		o = s.taboption('general', form.Value, 'api_host', _('API Host'));
+		o.default = '0.0.0.0';
+		o.placeholder = '0.0.0.0';
+		o.description = _('Use 0.0.0.0 to listen on all interfaces');
+
+		o = s.taboption('general', form.Value, 'memory_limit', _('Memory Limit'));
+		o.default = '2G';
+		o.placeholder = '2G';
+		o.description = _('Maximum memory for the container (e.g., 2G, 4G)');
+
+		o = s.taboption('general', form.Value, 'threads', _('CPU Threads'));
+		o.datatype = 'uinteger';
+		o.default = '4';
+		o.placeholder = '4';
+		o.description = _('Number of CPU threads for inference');
+
+		o = s.taboption('general', form.Value, 'context_size', _('Context Size'));
+		o.datatype = 'uinteger';
+		o.default = '2048';
+		o.placeholder = '2048';
+		o.description = _('Maximum context window size in tokens');
+
+		o = s.taboption('general', form.Flag, 'cors', _('Enable CORS'));
+		o.default = '1';
+		o.description = _('Allow cross-origin requests to API');
+
+		o = s.taboption('general', form.Flag, 'debug', _('Debug Mode'));
+		o.default = '0';
+		o.description = _('Enable verbose logging');
+
+		// Paths tab options
+		o = s.taboption('paths', form.Value, 'data_path', _('Data Directory'));
+		o.default = '/srv/localai';
+		o.placeholder = '/srv/localai';
+
+		o = s.taboption('paths', form.Value, 'models_path', _('Models Directory'));
+		o.default = '/srv/localai/models';
+		o.placeholder = '/srv/localai/models';
+
+		// GPU section (experimental)
+		s = m.section(form.TypedSection, 'gpu', _('GPU Acceleration (Experimental)'));
+		s.anonymous = true;
+		s.addremove = false;
+
+		o = s.option(form.Flag, 'enabled', _('Enable GPU'));
+		o.default = '0';
+		o.description = _('Enable GPU acceleration (requires compatible hardware)');
+
+		o = s.option(form.ListValue, 'backend', _('GPU Backend'));
+		o.value('vulkan', 'Vulkan (ARM64)');
+		o.value('cuda', 'CUDA (Nvidia)');
+		o.value('rocm', 'ROCm (AMD)');
+		o.default = 'vulkan';
+		o.depends('enabled', '1');
+
+		return m.render();
+	}
+});
diff --git a/package/secubox/luci-app-localai/root/usr/libexec/rpcd/luci.localai b/package/secubox/luci-app-localai/root/usr/libexec/rpcd/luci.localai
new file mode 100644
index 00000000..fc899dbd
--- /dev/null
+++ b/package/secubox/luci-app-localai/root/usr/libexec/rpcd/luci.localai
@@ -0,0 +1,445 @@
+#!/bin/sh
+# RPCD backend for LocalAI LuCI integration
+# Copyright (C) 2025 CyberMind.fr
+
+. /lib/functions.sh
+
+CONFIG="localai"
+LOCALAI_CTL="/usr/sbin/localaictl"
+
+# Load UCI config
+load_config() {
+	config_load "$CONFIG"
+	config_get API_PORT main api_port "8080"
+	config_get DATA_PATH main data_path "/srv/localai"
+	config_get MODELS_PATH main models_path "/srv/localai/models"
+	config_get MEMORY_LIMIT main memory_limit "2G"
+	config_get THREADS main threads "4"
+	config_get CONTEXT_SIZE main context_size "2048"
+}
+
+# Check if LocalAI is running
+is_running() {
+	pgrep local-ai >/dev/null 2>&1
+}
+
+# Get service status
+get_status() {
+	load_config
+	local running="false"
+	local uptime=0
+
+	if is_running; then
+		running="true"
+		# Get process uptime
+		local pid=$(pgrep local-ai | head -1)
+		if [ -n "$pid" ] && [ -d "/proc/$pid" ]; then
+			local start_time=$(stat -c %Y /proc/$pid 2>/dev/null || echo 0)
+			local now=$(date +%s)
+			uptime=$((now - start_time))
+		fi
+	fi
+
+	# Get enabled status
+	local enabled="false"
+	[ "$(uci -q get ${CONFIG}.main.enabled)" = "1" ] && enabled="true"
+
+	cat <<EOF
+{
+	"running": $running,
+	"enabled": $enabled,
+	"uptime": $uptime,
+	"api_port": $API_PORT,
+	"memory_limit": "$MEMORY_LIMIT",
+	"threads": $THREADS,
+	"context_size": $CONTEXT_SIZE,
+	"data_path": "$DATA_PATH",
+	"models_path": "$MODELS_PATH"
+}
+EOF
+}
+
+# Get installed models - queries both LocalAI API and filesystem
+get_models() {
+	load_config
+
+	local tmpfile="/tmp/localai_models_$$"
+	local first=1
+	local seen=""
+
+	echo '{"models":['
+
+	# First, try to get models from LocalAI API (shows loaded/active models)
+	if is_running; then
+		wget -q -O "$tmpfile" "http://127.0.0.1:$API_PORT/v1/models" 2>/dev/null
+		if [ -f "$tmpfile" ] && [ -s "$tmpfile" ]; then
+			# Try indexed access for each model (max 20)
+			local i=0
+			while [ $i -lt 20 ]; do
+				local model_id=$(jsonfilter -i "$tmpfile" -e "@.data[$i].id" 2>/dev/null)
+				[ -z "$model_id" ] && break
+				[ $first -eq 0 ] && echo ","
+				first=0
+				seen="$seen $model_id"
+				cat <<EOF
+{
+	"id": "$model_id",
+	"name": "$model_id",
+	"size": 0,
+	"type": "loaded",
+	"loaded": true
+}
+EOF
+				i=$((i + 1))
+			done
+		fi
+		rm -f "$tmpfile"
+	fi
+
+	# Scan filesystem for model files
+	if [ -d "$MODELS_PATH" ]; then
+		for model in "$MODELS_PATH"/*.gguf "$MODELS_PATH"/*.bin "$MODELS_PATH"/*.onnx; do
+			[ -f "$model" ] || continue
+			local name=$(basename "$model")
+			local basename_no_ext="${name%.*}"
+			local size=$(stat -c %s "$model" 2>/dev/null || echo 0)
+			local ext="${name##*.}"
+			local type="unknown"
+			local loaded="false"
+
+			case "$ext" in
+				gguf) type="llama-cpp" ;;
+				bin) type="transformers" ;;
+				onnx) type="onnx" ;;
+			esac
+
+			# Check if this model is in the seen list (loaded from API)
+			case " $seen " in
+				*" $basename_no_ext "*) continue ;;
+			esac
+
+			[ $first -eq 0 ] && echo ","
+			first=0
+
+			cat <<EOF
+{
+	"id": "$basename_no_ext",
+	"name": "$name",
+	"size": $size,
+	"type": "$type",
+	"path": "$model",
+	"loaded": $loaded
+}
+EOF
+		done
+	fi
+
+	# Also scan for YAML model configs (LocalAI model definitions)
+	if [ -d "$MODELS_PATH" ]; then
+		for yaml in "$MODELS_PATH"/*.yaml "$MODELS_PATH"/*.yml; do
+			[ -f "$yaml" ] || continue
+			local name=$(basename "$yaml")
+			local basename_no_ext="${name%.*}"
+
+			# Skip if already seen
+			case " $seen " in
+				*" $basename_no_ext "*) continue ;;
+			esac
+
+			# Check if there's a model file with same base name already shown
+			local already_shown=0
+			for ext in gguf bin onnx; do
+				[ -f "$MODELS_PATH/$basename_no_ext.$ext" ] && already_shown=1
+			done
+			[ $already_shown -eq 1 ] && continue
+
+			[ $first -eq 0 ] && echo ","
+			first=0
+
+			cat <<EOF
+{
+	"id": "$basename_no_ext",
+	"name": "$basename_no_ext",
+	"size": 0,
+	"type": "config",
+	"path": "$yaml",
+	"loaded": false
+}
+EOF
+		done
+	fi
+
+	echo ']}'
+}
+
+# Get configuration
+get_config() {
+	load_config
+
+	cat <<EOF
+{
+	"api_port": $API_PORT,
+	"data_path": "$DATA_PATH",
+	"models_path": "$MODELS_PATH",
+	"memory_limit": "$MEMORY_LIMIT",
+	"threads": $THREADS,
+	"context_size": $CONTEXT_SIZE
+}
+EOF
+}
+
+# Health check
+get_health() {
+	load_config
+
+	local healthy="false"
+	local api_status="unknown"
+
+	if is_running; then
+		# Check API health endpoint
+		local response=$(wget -q -O - "http://127.0.0.1:$API_PORT/readyz" 2>/dev/null)
+		if echo "$response" | grep -q "ok"; then
+			healthy="true"
+			api_status="ok"
+		else
+			api_status="unhealthy"
+		fi
+	else
+		api_status="stopped"
+	fi
+
+	cat <<EOF
+{
+	"healthy": $healthy,
+	"api_status": "$api_status"
+}
+EOF
+}
+
+# Get metrics
+get_metrics() {
+	load_config
+
+	local mem_used=0
+	local cpu_percent=0
+
+	if is_running; then
+		local pid=$(pgrep local-ai | head -1)
+		if [ -n "$pid" ]; then
+			# Get memory usage from /proc
+			mem_used=$(awk '/VmRSS/ {print $2*1024}' /proc/$pid/status 2>/dev/null || echo 0)
+
+			# Get CPU from ps
+			cpu_percent=$(ps -o %cpu= -p $pid 2>/dev/null | tr -d ' ' || echo "0")
+		fi
+	fi
+
+	cat <<EOF
+{
+	"memory_used": $mem_used,
+	"cpu_percent": $cpu_percent
+}
+EOF
+}
+
+# Start service
+do_start() {
+	if is_running; then
+		echo '{"success":false,"error":"Already running"}'
+		return
+	fi
+
+	/etc/init.d/localai start >/dev/null 2>&1
+	sleep 2
+
+	if is_running; then
+		echo '{"success":true}'
+	else
+		echo '{"success":false,"error":"Failed to start"}'
+	fi
+}
+
+# Stop service
+do_stop() {
+	/etc/init.d/localai stop >/dev/null 2>&1
+	sleep 1
+
+	if ! is_running; then
+		echo '{"success":true}'
+	else
+		echo '{"success":false,"error":"Failed to stop"}'
+	fi
+}
+
+# Restart service
+do_restart() {
+	/etc/init.d/localai restart >/dev/null 2>&1
+	sleep 3
+
+	if is_running; then
+		echo '{"success":true}'
+	else
+		echo '{"success":false,"error":"Failed to restart"}'
+	fi
+}
+
+# Install model
+do_model_install() {
+	local name="$1"
+	[ -z "$name" ] && { echo '{"success":false,"error":"Model name required"}'; return; }
+
+	local output=$($LOCALAI_CTL model-install "$name" 2>&1)
+	local ret=$?
+
+	if [ $ret -eq 0 ]; then
+		echo '{"success":true}'
+	else
+		local error=$(echo "$output" | tail -1 | sed 's/"/\\"/g')
+		echo "{\"success\":false,\"error\":\"$error\"}"
+	fi
+}
+
+# Remove model
+do_model_remove() {
+	local name="$1"
+	[ -z "$name" ] && { echo '{"success":false,"error":"Model name required"}'; return; }
+
+	local output=$($LOCALAI_CTL model-remove "$name" 2>&1)
+	local ret=$?
+
+	if [ $ret -eq 0 ]; then
+		echo '{"success":true}'
+	else
+		local error=$(echo "$output" | tail -1 | sed 's/"/\\"/g')
+		echo "{\"success\":false,\"error\":\"$error\"}"
+	fi
+}
+
+# Chat completion (proxy to LocalAI API)
+do_chat() {
+	load_config
+	local model="$1"
+	local messages="$2"
+
+	if ! is_running; then
+		echo '{"response":"","error":"LocalAI is not running"}'
+		return
+	fi
+
+	# Validate inputs
+	[ -z "$model" ] && { echo '{"response":"","error":"Model not specified"}'; return; }
+	[ -z "$messages" ] && { echo '{"response":"","error":"Messages not provided"}'; return; }
+
+	# Build request body - messages should already be a JSON array string
+	local request_body="{\"model\":\"$model\",\"messages\":$messages}"
+
+	# Call LocalAI API using a temp file for better handling
+	local tmpfile="/tmp/localai_chat_$$"
+	local http_code
+
+	http_code=$(wget -q -O "$tmpfile" --post-data "$request_body" \
+		--header="Content-Type: application/json" \
+		"http://127.0.0.1:$API_PORT/v1/chat/completions" 2>/dev/null; echo $?)
+
+	if [ -f "$tmpfile" ] && [ -s "$tmpfile" ]; then
+		# Extract message content using jsonfilter
+		local content=$(jsonfilter -i "$tmpfile" -e '@.choices[0].message.content' 2>/dev/null)
+		local error=$(jsonfilter -i "$tmpfile" -e '@.error.message' 2>/dev/null)
+		rm -f "$tmpfile"
+
+		if [ -n "$error" ]; then
+			# Escape quotes and newlines in error
+			error=$(echo "$error" | sed 's/"/\\"/g' | tr '\n' ' ')
+			echo "{\"response\":\"\",\"error\":\"$error\"}"
+		elif [ -n "$content" ]; then
+			# Escape quotes and newlines in content
+			content=$(echo "$content" | sed 's/"/\\"/g' | tr '\n' '\\n')
+			echo "{\"response\":\"$content\"}"
+		else
+			echo '{"response":"","error":"Empty response from API"}'
+		fi
+	else
+		rm -f "$tmpfile" 2>/dev/null
+		echo '{"response":"","error":"API request failed - no response"}'
+	fi
+}
+
+# Text completion
+do_complete() {
+	load_config
+	local model="$1"
+	local prompt="$2"
+
+	if ! is_running; then
+		echo '{"text":"","error":"LocalAI is not running"}'
+		return
+	fi
+
+	local response=$(wget -q -O - --post-data "{\"model\":\"$model\",\"prompt\":\"$prompt\"}" \
+		--header="Content-Type: application/json" \
+		"http://127.0.0.1:$API_PORT/v1/completions" 2>/dev/null)
+
+	if [ -n "$response" ]; then
+		local text=$(echo "$response" | jsonfilter -e '@.choices[0].text' 2>/dev/null)
+		echo "{\"text\":\"$(echo "$text" | sed 's/"/\\"/g')\"}"
+	else
+		echo '{"text":"","error":"API request failed"}'
+	fi
+}
+
+# UBUS method list
+case "$1" in
+	list)
+		cat <<'EOF'
+{
+	"status": {},
+	"models": {},
+	"config": {},
+	"health": {},
+	"metrics": {},
+	"start": {},
+	"stop": {},
+	"restart": {},
+	"model_install": {"name": "string"},
+	"model_remove": {"name": "string"},
+	"chat": {"model": "string", "messages": "array"},
+	"complete": {"model": "string", "prompt": "string"}
+}
+EOF
+		;;
+	call)
+		case "$2" in
+			status) get_status ;;
+			models) get_models ;;
+			config) get_config ;;
+			health) get_health ;;
+			metrics) get_metrics ;;
+			start) do_start ;;
+			stop) do_stop ;;
+			restart) do_restart ;;
+			model_install)
+				read -r input
+				name=$(echo "$input" | jsonfilter -e '@.name' 2>/dev/null)
+				do_model_install "$name"
+				;;
+			model_remove)
+				read -r input
+				name=$(echo "$input" | jsonfilter -e '@.name' 2>/dev/null)
+				do_model_remove "$name"
+				;;
+			chat)
+				read -r input
+				model=$(echo "$input" | jsonfilter -e '@.model' 2>/dev/null)
+				messages=$(echo "$input" | jsonfilter -e '@.messages' 2>/dev/null)
+				do_chat "$model" "$messages"
+				;;
+			complete)
+				read -r input
+				model=$(echo "$input" | jsonfilter -e '@.model' 2>/dev/null)
+				prompt=$(echo "$input" | jsonfilter -e '@.prompt' 2>/dev/null)
+				do_complete "$model" "$prompt"
+				;;
+			*) echo '{"error":"Unknown method"}' ;;
+		esac
+		;;
+esac
diff --git a/package/secubox/luci-app-localai/root/usr/share/luci/menu.d/luci-app-localai.json b/package/secubox/luci-app-localai/root/usr/share/luci/menu.d/luci-app-localai.json
new file mode 100644
index 00000000..d80a0f44
--- /dev/null
+++ b/package/secubox/luci-app-localai/root/usr/share/luci/menu.d/luci-app-localai.json
@@ -0,0 +1,56 @@
+{
+	"admin/secubox/services/localai": {
+		"title": "LocalAI",
+		"order": 50,
+		"action": {
+			"type": "firstchild"
+		},
+		"depends": {
+			"acl": ["luci-app-localai"]
+		}
+	},
+	"admin/secubox/services/localai/dashboard": {
+		"title": "Dashboard",
+		"order": 10,
+		"action": {
+			"type": "view",
+			"path": "localai/dashboard"
+		},
+		"depends": {
+			"acl": ["luci-app-localai"]
+		}
+	},
+	"admin/secubox/services/localai/models": {
+		"title": "Models",
+		"order": 20,
+		"action": {
+			"type": "view",
+			"path": "localai/models"
+		},
+		"depends": {
+			"acl": ["luci-app-localai"]
+		}
+	},
+	"admin/secubox/services/localai/chat": {
+		"title": "Chat",
+		"order": 30,
+		"action": {
+			"type": "view",
+			"path": "localai/chat"
+		},
+		"depends": {
+			"acl": ["luci-app-localai"]
+		}
+	},
+	"admin/secubox/services/localai/settings": {
+		"title": "Settings",
+		"order": 90,
+		"action": {
+			"type": "view",
+			"path": "localai/settings"
+		},
+		"depends": {
+			"acl": ["luci-app-localai"]
+		}
+	}
+}
diff --git a/package/secubox/luci-app-localai/root/usr/share/rpcd/acl.d/luci-app-localai.json b/package/secubox/luci-app-localai/root/usr/share/rpcd/acl.d/luci-app-localai.json
new file mode 100644
index 00000000..3fda8466
--- /dev/null
+++ b/package/secubox/luci-app-localai/root/usr/share/rpcd/acl.d/luci-app-localai.json
@@ -0,0 +1,39 @@
+{
+	"luci-app-localai": {
+		"description": "Grant access to LuCI LocalAI Dashboard",
+		"read": {
+			"ubus": {
+				"luci.localai": [
+					"status",
+					"models",
+					"config",
+					"health",
+					"metrics"
+				],
+				"system": [ "info", "board" ],
+				"file": [ "read", "stat", "exec" ]
+			},
+			"uci": [ "localai" ],
+			"file": {
+				"/etc/config/localai": [ "read" ],
+				"/srv/localai/*": [ "read" ]
+			}
+		},
+		"write": {
+			"ubus": {
+				"luci.localai": [
+					"start",
+					"stop",
+					"restart",
+					"model_install",
+					"model_remove",
+					"model_load",
+					"model_unload",
+					"chat",
+					"complete"
+				]
+			},
+			"uci": [ "localai" ]
+		}
+	}
+}
diff --git a/package/secubox/luci-app-secubox-portal/Makefile b/package/secubox/luci-app-secubox-portal/Makefile
index 54f7e26e..aea428d0 100644
--- a/package/secubox/luci-app-secubox-portal/Makefile
+++ b/package/secubox/luci-app-secubox-portal/Makefile
@@ -11,7 +11,7 @@ LUCI_DESCRIPTION:=Unified entry point for all SecuBox applications with tabbed n
 LUCI_DEPENDS:=+luci-base +luci-theme-secubox
 LUCI_PKGARCH:=all
 PKG_VERSION:=0.6.0
-PKG_RELEASE:=2
+PKG_RELEASE:=5
 PKG_LICENSE:=GPL-3.0-or-later
 PKG_MAINTAINER:=SecuBox Team <secubox@example.com>
 
diff --git a/package/secubox/luci-app-secubox-portal/htdocs/luci-static/resources/secubox-portal/portal.js b/package/secubox/luci-app-secubox-portal/htdocs/luci-static/resources/secubox-portal/portal.js
index 76c0ad33..010d16cc 100644
--- a/package/secubox/luci-app-secubox-portal/htdocs/luci-static/resources/secubox-portal/portal.js
+++ b/package/secubox/luci-app-secubox-portal/htdocs/luci-static/resources/secubox-portal/portal.js
@@ -235,6 +235,30 @@ return baseclass.extend({
 			path: 'admin/secubox/services/mmpm/dashboard',
 			service: 'mmpm',
 			version: '3.1.0'
+		},
+		'glances': {
+			id: 'glances',
+			name: 'Glances',
+			desc: 'Cross-platform system monitoring tool with web interface',
+			icon: '\ud83d\udcca',
+			iconBg: 'rgba(16, 185, 129, 0.15)',
+			iconColor: '#10b981',
+			section: 'monitoring',
+			path: 'admin/secubox/monitoring/glances/dashboard',
+			service: 'glances',
+			version: '4.2.1'
+		},
+		'localai': {
+			id: 'localai',
+			name: 'LocalAI',
+			desc: 'Self-hosted, privacy-first AI/LLM with OpenAI-compatible API',
+			icon: '\ud83e\udd16',
+			iconBg: 'rgba(168, 85, 247, 0.15)',
+			iconColor: '#a855f7',
+			section: 'services',
+			path: 'admin/secubox/services/localai/dashboard',
+			service: 'localai',
+			version: '3.10.0'
 		}
 	},
 
diff --git a/package/secubox/luci-app-secubox-portal/root/usr/share/luci/menu.d/luci-app-secubox-portal.json b/package/secubox/luci-app-secubox-portal/root/usr/share/luci/menu.d/luci-app-secubox-portal.json
index 1d56a7ad..ed653236 100644
--- a/package/secubox/luci-app-secubox-portal/root/usr/share/luci/menu.d/luci-app-secubox-portal.json
+++ b/package/secubox/luci-app-secubox-portal/root/usr/share/luci/menu.d/luci-app-secubox-portal.json
@@ -21,6 +21,16 @@
 			"acl": ["luci-app-secubox-portal"]
 		}
 	},
+	"admin/secubox/services": {
+		"title": "Services",
+		"order": 50,
+		"action": {
+			"type": "firstchild"
+		},
+		"depends": {
+			"acl": ["luci-app-secubox-portal"]
+		}
+	},
 	"admin/secubox/apps": {
 		"title": "Apps",
 		"order": 60,
diff --git a/package/secubox/secubox-app-localai/Makefile b/package/secubox/secubox-app-localai/Makefile
new file mode 100644
index 00000000..388c5609
--- /dev/null
+++ b/package/secubox/secubox-app-localai/Makefile
@@ -0,0 +1,73 @@
+include $(TOPDIR)/rules.mk
+
+PKG_NAME:=secubox-app-localai
+PKG_RELEASE:=4
+PKG_VERSION:=0.1.0
+PKG_ARCH:=all
+PKG_MAINTAINER:=CyberMind Studio <contact@cybermind.fr>
+PKG_LICENSE:=Apache-2.0
+
+include $(INCLUDE_DIR)/package.mk
+
+define Package/secubox-app-localai
+  SECTION:=utils
+  CATEGORY:=Utilities
+  PKGARCH:=all
+  SUBMENU:=SecuBox Apps
+  TITLE:=SecuBox LocalAI - Self-hosted LLM (LXC)
+  DEPENDS:=+uci +libuci +jsonfilter
+endef
+
+define Package/secubox-app-localai/description
+LocalAI - Self-hosted, privacy-first AI/LLM for SecuBox-powered OpenWrt systems.
+
+Features:
+- OpenAI-compatible API (drop-in replacement)
+- No cloud dependency - all processing on-device
+- Support for various models (LLaMA, Mistral, Phi, etc.)
+- Text generation, embeddings, transcription
+- Image generation (optional)
+
+Runs in LXC container for isolation.
+Configure in /etc/config/localai.
+endef
+
+define Package/secubox-app-localai/conffiles
+/etc/config/localai
+endef
+
+define Build/Compile
+endef
+
+define Package/secubox-app-localai/install
+	$(INSTALL_DIR) $(1)/etc/config
+	$(INSTALL_CONF) ./files/etc/config/localai $(1)/etc/config/localai
+
+	$(INSTALL_DIR) $(1)/etc/init.d
+	$(INSTALL_BIN) ./files/etc/init.d/localai $(1)/etc/init.d/localai
+
+	$(INSTALL_DIR) $(1)/usr/sbin
+	$(INSTALL_BIN) ./files/usr/sbin/localaictl $(1)/usr/sbin/localaictl
+endef
+
+define Package/secubox-app-localai/postinst
+#!/bin/sh
+[ -n "$${IPKG_INSTROOT}" ] || {
+	echo ""
+	echo "LocalAI installed."
+	echo ""
+	echo "To install and start LocalAI:"
+	echo "  localaictl install"
+	echo "  /etc/init.d/localai start"
+	echo ""
+	echo "API endpoint: http://<router-ip>:8080/v1"
+	echo "Web UI: http://<router-ip>:8080"
+	echo ""
+	echo "Download models with:"
+	echo "  localaictl model-install <model-name>"
+	echo ""
+}
+exit 0
+endef
+
+$(eval $(call BuildPackage,secubox-app-localai))
diff --git a/package/secubox/secubox-app-localai/files/etc/config/localai b/package/secubox/secubox-app-localai/files/etc/config/localai
new file mode 100644
index 00000000..8eb4af4b
--- /dev/null
+++ b/package/secubox/secubox-app-localai/files/etc/config/localai
@@ -0,0 +1,51 @@
+config main 'main'
+	option enabled '0'
+	option api_port '8080'
+	option api_host '0.0.0.0'
+	option data_path '/srv/localai'
+	option models_path '/srv/localai/models'
+	option memory_limit '2G'
+	option threads '4'
+	option context_size '2048'
+	option debug '0'
+	option cors '1'
+
+# GPU settings (experimental on ARM64)
+config gpu 'gpu'
+	option enabled '0'
+	option backend 'vulkan'
+
+# Default model to load on startup
+config model 'default'
+	option enabled '1'
+	option name 'phi-2'
+	option backend 'llama-cpp'
+
+# Model presets
+config preset 'phi2'
+	option name 'phi-2'
+	option url 'https://huggingface.co/TheBloke/phi-2-GGUF/resolve/main/phi-2.Q4_K_M.gguf'
+	option size '1.6G'
+	option type 'text-generation'
+	option description 'Microsoft Phi-2 - Compact and efficient'
+
+config preset 'mistral'
+	option name 'mistral-7b'
+	option url 'https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.2-GGUF/resolve/main/mistral-7b-instruct-v0.2.Q4_K_M.gguf'
+	option size '4.1G'
+	option type 'text-generation'
+	option description 'Mistral 7B Instruct - High quality assistant'
+
+config preset 'tinyllama'
+	option name 'tinyllama'
+	option url 'https://huggingface.co/TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF/resolve/main/tinyllama-1.1b-chat-v1.0.Q4_K_M.gguf'
+	option size '669M'
+	option type 'text-generation'
+	option description 'TinyLlama 1.1B - Ultra-lightweight'
+
+config preset 'gte_small'
+	option name 'gte-small'
+	option url 'https://huggingface.co/Supabase/gte-small/resolve/main/model.onnx'
+	option size '67M'
+	option type 'embeddings'
+	option description 'GTE Small - Fast embeddings'
diff --git a/package/secubox/secubox-app-localai/files/etc/init.d/localai b/package/secubox/secubox-app-localai/files/etc/init.d/localai
new file mode 100644
index 00000000..239c00a5
--- /dev/null
+++ b/package/secubox/secubox-app-localai/files/etc/init.d/localai
@@ -0,0 +1,40 @@
+#!/bin/sh /etc/rc.common
+# SecuBox LocalAI - Self-hosted LLM service
+# Copyright (C) 2025 CyberMind.fr
+
+START=95
+STOP=10
+USE_PROCD=1
+
+PROG=/usr/sbin/localaictl
+
+start_service() {
+	local enabled
+	config_load localai
+	config_get enabled main enabled '0'
+
+	[ "$enabled" = "1" ] || {
+		echo "LocalAI is disabled. Enable with: uci set localai.main.enabled=1"
+		return 0
+	}
+
+	procd_open_instance
+	procd_set_param command $PROG service-run
+	procd_set_param respawn 3600 5 5
+	procd_set_param stdout 1
+	procd_set_param stderr 1
+	procd_close_instance
+}
+
+stop_service() {
+	$PROG service-stop
+}
+
+service_triggers() {
+	procd_add_reload_trigger "localai"
+}
+
+reload_service() {
+	stop
+	start
+}
diff --git a/package/secubox/secubox-app-localai/files/usr/sbin/localaictl b/package/secubox/secubox-app-localai/files/usr/sbin/localaictl
new file mode 100644
index 00000000..62ec8a26
--- /dev/null
+++ b/package/secubox/secubox-app-localai/files/usr/sbin/localaictl
@@ -0,0 +1,578 @@
+#!/bin/sh
+# SecuBox LocalAI manager - LXC container support
+# Copyright (C) 2025 CyberMind.fr
+
+CONFIG="localai"
+LXC_NAME="localai"
+OPKG_UPDATED=0
+LOCALAI_VERSION="v3.10.0"
+
+# Paths
+LXC_PATH="/srv/lxc"
+LXC_ROOTFS="$LXC_PATH/$LXC_NAME/rootfs"
+LXC_CONFIG="$LXC_PATH/$LXC_NAME/config"
+
+usage() {
+	cat <<'EOF'
+Usage: localaictl <command>
+
+Commands:
+  install            Install prerequisites and create LXC container
+  check              Run prerequisite checks
+  update             Update LocalAI in container
+  status             Show container and service status
+  logs               Show LocalAI logs (use -f to follow)
+  shell              Open shell in container
+
+Model Management:
+  models             List installed models
+  model-install <n>  Install model from preset or URL
+  model-remove <n>   Remove installed model
+  model-load <n>     Load model into memory
+  model-unload <n>   Unload model from memory
+
+Service Control:
+  service-run        Internal: run container under procd
+  service-stop       Stop container
+
+API Endpoints (default port 8080):
+  /v1/chat/completions   - Chat completion (OpenAI compatible)
+  /v1/completions        - Text completion
+  /v1/embeddings         - Generate embeddings
+  /v1/models             - List available models
+  /                      - Web UI
+
+Configuration: /etc/config/localai
+EOF
+}
+
+require_root() { [ "$(id -u)" -eq 0 ] || { echo "Root required" >&2; exit 1; }; }
+
+log_info() { echo "[INFO] $*"; }
+log_warn() { echo "[WARN] $*" >&2; }
+log_error() { echo "[ERROR] $*" >&2; }
+
+uci_get() { uci -q get ${CONFIG}.$1; }
+uci_set() { uci set ${CONFIG}.$1="$2" && uci commit ${CONFIG}; }
+uci_get_list() { uci -q get ${CONFIG}.$1 2>/dev/null; }
+
+# Load configuration with defaults
+load_config() {
+	api_port="$(uci_get main.api_port || echo 8080)"
+	api_host="$(uci_get main.api_host || echo 0.0.0.0)"
+	data_path="$(uci_get main.data_path || echo /srv/localai)"
+	models_path="$(uci_get main.models_path || echo /srv/localai/models)"
+	memory_limit="$(uci_get main.memory_limit || echo 2G)"
+	threads="$(uci_get main.threads || echo 4)"
+	context_size="$(uci_get main.context_size || echo 2048)"
+	debug="$(uci_get main.debug || echo 0)"
+	cors="$(uci_get main.cors || echo 1)"
+	gpu_enabled="$(uci_get gpu.enabled || echo 0)"
+	gpu_backend="$(uci_get gpu.backend || echo vulkan)"
+}
+
+ensure_dir() { [ -d "$1" ] || mkdir -p "$1"; }
+
+has_lxc() {
+	command -v lxc-start >/dev/null 2>&1 && \
+	command -v lxc-stop >/dev/null 2>&1
+}
+
+# Ensure required packages are installed
+ensure_packages() {
+	require_root
+	for pkg in "$@"; do
+		if ! opkg list-installed | grep -q "^$pkg "; then
+			if [ "$OPKG_UPDATED" -eq 0 ]; then
+				opkg update || return 1
+				OPKG_UPDATED=1
+			fi
+			opkg install "$pkg" || return 1
+		fi
+	done
+}
+
+# =============================================================================
+# LXC CONTAINER FUNCTIONS
+# =============================================================================
+
+lxc_check_prereqs() {
+	log_info "Checking LXC prerequisites..."
+	ensure_packages lxc lxc-common lxc-attach lxc-start lxc-stop lxc-destroy || return 1
+
+	if [ ! -d /sys/fs/cgroup ]; then
+		log_error "cgroups not mounted at /sys/fs/cgroup"
+		return 1
+	fi
+
+	log_info "LXC ready"
+}
+
+lxc_create_rootfs() {
+	load_config
+
+	if [ -d "$LXC_ROOTFS" ] && [ -x "$LXC_ROOTFS/usr/bin/local-ai" ]; then
+		log_info "LXC rootfs already exists with LocalAI"
+		return 0
+	fi
+
+	log_info "Creating LXC rootfs for LocalAI..."
+	ensure_dir "$LXC_PATH/$LXC_NAME"
+
+	lxc_download_binary || return 1
+	lxc_create_config || return 1
+
+	log_info "LXC rootfs created successfully"
+}
+
+lxc_download_binary() {
+	local rootfs="$LXC_ROOTFS"
+	local arch
+
+	# Detect architecture - LocalAI uses lowercase format: local-ai-vX.X.X-linux-arm64
+	case "$(uname -m)" in
+		x86_64) arch="linux-x86_64" ;;
+		aarch64) arch="linux-arm64" ;;
+		armv7l) arch="linux-arm" ;;
+		*) arch="linux-x86_64" ;;
+	esac
+
+	log_info "Downloading LocalAI $LOCALAI_VERSION for $arch..."
+	ensure_dir "$rootfs/usr/bin"
+	ensure_dir "$rootfs/data"
+	ensure_dir "$rootfs/models"
+	ensure_dir "$rootfs/tmp"
+	ensure_dir "$rootfs/etc"
+
+	# Download LocalAI binary - format: local-ai-v3.10.0-linux-arm64
+	local binary_url="https://github.com/mudler/LocalAI/releases/download/${LOCALAI_VERSION}/local-ai-${LOCALAI_VERSION}-${arch}"
+
+	log_info "Downloading from: $binary_url"
+	if wget -q --show-progress -O "$rootfs/usr/bin/local-ai" "$binary_url"; then
+		chmod +x "$rootfs/usr/bin/local-ai"
+		log_info "LocalAI binary downloaded successfully ($(ls -sh "$rootfs/usr/bin/local-ai" | cut -d' ' -f1))"
+	else
+		log_error "Failed to download LocalAI binary"
+		log_error "URL: $binary_url"
+		return 1
+	fi
+
+	# Create minimal rootfs structure
+	mkdir -p "$rootfs/bin" "$rootfs/lib" "$rootfs/proc" "$rootfs/sys" "$rootfs/dev"
+
+	# Create resolv.conf
+	echo "nameserver 8.8.8.8" > "$rootfs/etc/resolv.conf"
+
+	# Create startup script
+	cat > "$rootfs/usr/bin/start-localai.sh" << 'START'
+#!/bin/sh
+export PATH="/usr/bin:/bin:$PATH"
+cd /data
+
+# Read environment variables
+API_PORT="${LOCALAI_API_PORT:-8080}"
+API_HOST="${LOCALAI_API_HOST:-0.0.0.0}"
+THREADS="${LOCALAI_THREADS:-4}"
+CONTEXT_SIZE="${LOCALAI_CONTEXT_SIZE:-2048}"
+DEBUG="${LOCALAI_DEBUG:-0}"
+CORS="${LOCALAI_CORS:-1}"
+GPU_ENABLED="${LOCALAI_GPU_ENABLED:-0}"
+
+# Build args
+ARGS="--address ${API_HOST}:${API_PORT}"
+ARGS="$ARGS --models-path /models"
+ARGS="$ARGS --threads $THREADS"
+ARGS="$ARGS --context-size $CONTEXT_SIZE"
+
+[ "$DEBUG" = "1" ] && ARGS="$ARGS --debug"
+[ "$CORS" = "1" ] && ARGS="$ARGS --cors"
+
+echo "Starting LocalAI..."
+echo "API: http://${API_HOST}:${API_PORT}"
+echo "Models path: /models"
+echo "Threads: $THREADS, Context: $CONTEXT_SIZE"
+
+exec /usr/bin/local-ai $ARGS
+START
+	chmod +x "$rootfs/usr/bin/start-localai.sh"
+
+	log_info "LocalAI binary and startup script installed"
+}
+
+lxc_create_config() {
+	load_config
+
+	# Build command line flags
+	local cors_flag=""
+	local debug_flag=""
+	[ "$cors" = "1" ] && cors_flag=" --cors"
+	[ "$debug" = "1" ] && debug_flag=" --debug"
+
+	cat > "$LXC_CONFIG" << EOF
+# LocalAI LXC Configuration
+lxc.uts.name = $LXC_NAME
+
+# Root filesystem
+lxc.rootfs.path = dir:$LXC_ROOTFS
+
+# Network - use host network for simplicity
+lxc.net.0.type = none
+
+# Mounts
+lxc.mount.auto = proc:mixed sys:ro cgroup:mixed
+lxc.mount.entry = $data_path data none bind,create=dir 0 0
+lxc.mount.entry = $models_path models none bind,create=dir 0 0
+
+# Environment variables for configuration
+lxc.environment = LOCALAI_API_PORT=$api_port
+lxc.environment = LOCALAI_API_HOST=$api_host
+lxc.environment = LOCALAI_THREADS=$threads
+lxc.environment = LOCALAI_CONTEXT_SIZE=$context_size
+lxc.environment = LOCALAI_DEBUG=$debug
+lxc.environment = LOCALAI_CORS=$cors
+lxc.environment = LOCALAI_GPU_ENABLED=$gpu_enabled
+
+# Capabilities
+lxc.cap.drop = sys_admin sys_module mac_admin mac_override
+
+# cgroups limits
+lxc.cgroup.memory.limit_in_bytes = $memory_limit
+
+# Run binary directly (no shell needed in minimal rootfs)
+lxc.init.cmd = /usr/bin/local-ai --address ${api_host}:${api_port} --models-path /models --threads $threads --context-size $context_size${cors_flag}${debug_flag}
+
+# Console
+lxc.console.size = 4096
+lxc.pty.max = 1024
+EOF
+
+	log_info "LXC config created at $LXC_CONFIG"
+}
+
+lxc_stop() {
+	if lxc-info -n "$LXC_NAME" >/dev/null 2>&1; then
+		lxc-stop -n "$LXC_NAME" -k >/dev/null 2>&1 || true
+	fi
+}
+
+lxc_run() {
+	load_config
+	lxc_stop
+
+	if [ ! -f "$LXC_CONFIG" ]; then
+		log_error "LXC not configured. Run 'localaictl install' first."
+		return 1
+	fi
+
+	# Regenerate config to pick up any UCI changes
+	lxc_create_config
+
+	# Ensure mount points exist
+	ensure_dir "$data_path"
+	ensure_dir "$models_path"
+
+	log_info "Starting LocalAI LXC container..."
+	log_info "API endpoint: http://0.0.0.0:$api_port/v1"
+	log_info "Web UI: http://0.0.0.0:$api_port"
+	log_info "Models path: $models_path"
+	exec lxc-start -n "$LXC_NAME" -F -f "$LXC_CONFIG"
+}
+
+lxc_status() {
+	load_config
+	echo "=== LocalAI Status ==="
+	echo ""
+
+	if lxc-info -n "$LXC_NAME" >/dev/null 2>&1; then
+		lxc-info -n "$LXC_NAME"
+	else
+		echo "LXC container '$LXC_NAME' not found or not configured"
+	fi
+
+	echo ""
+	echo "=== Configuration ==="
+	echo "API port: $api_port"
+	echo "Data path: $data_path"
+	echo "Models path: $models_path"
+	echo "Memory limit: $memory_limit"
+	echo "Threads: $threads"
+	echo "Context size: $context_size"
+	echo ""
+
+	# Check API health
+	if wget -q -O - "http://127.0.0.1:$api_port/readyz" 2>/dev/null | grep -q "ok"; then
+		echo "API Status: HEALTHY"
+	else
+		echo "API Status: NOT RESPONDING"
+	fi
+}
+
+lxc_logs() {
+	if [ "$1" = "-f" ]; then
+		logread -f -e localai
+	else
+		logread -e localai | tail -100
+	fi
+}
+
+lxc_shell() {
+	lxc-attach -n "$LXC_NAME" -- /bin/sh
+}
+
+lxc_destroy() {
+	lxc_stop
+	if [ -d "$LXC_PATH/$LXC_NAME" ]; then
+		rm -rf "$LXC_PATH/$LXC_NAME"
+		log_info "LXC container destroyed"
+	fi
+}
+
+# =============================================================================
+# MODEL MANAGEMENT
+# =============================================================================
+
+cmd_models() {
+	load_config
+	echo "=== Installed Models ==="
+	echo ""
+
+	if [ -d "$models_path" ]; then
+		local count=0
+		for model in "$models_path"/*.gguf "$models_path"/*.bin "$models_path"/*.onnx; do
+			[ -f "$model" ] || continue
+			count=$((count + 1))
+			local name=$(basename "$model")
+			local size=$(ls -lh "$model" | awk '{print $5}')
+			echo "  $count. $name ($size)"
+		done
+
+		if [ "$count" -eq 0 ]; then
+			echo "  No models installed"
+			echo ""
+			echo "Install a model with:"
+			echo "  localaictl model-install phi2"
+			echo "  localaictl model-install tinyllama"
+		fi
+	else
+		echo "  Models directory not found: $models_path"
+	fi
+
+	echo ""
+	echo "=== Available Presets ==="
+	echo ""
+
+	# List presets from UCI config
+	uci show localai 2>/dev/null | grep "preset\[" | while read line; do
+		local section=$(echo "$line" | cut -d. -f2 | cut -d= -f1)
+		local name=$(uci_get "$section.name")
+		local desc=$(uci_get "$section.description")
+		local size=$(uci_get "$section.size")
+		[ -n "$name" ] && echo "  $name - $desc ($size)"
+	done
+}
+
+cmd_model_install() {
+	load_config
+	require_root
+
+	local model_name="$1"
+	[ -z "$model_name" ] && { echo "Usage: localaictl model-install <preset-name|url>"; return 1; }
+
+	ensure_dir "$models_path"
+
+	# Check if it's a preset
+	local preset_url=""
+	local preset_file=""
+
+	# Search presets
+	for section in $(uci show localai 2>/dev/null | grep "=preset" | cut -d. -f2 | cut -d= -f1); do
+		local pname=$(uci_get "$section.name")
+		if [ "$pname" = "$model_name" ]; then
+			preset_url=$(uci_get "$section.url")
+			preset_file=$(basename "$preset_url")
+			break
+		fi
+	done
+
+	if [ -n "$preset_url" ]; then
+		log_info "Installing preset model: $model_name"
+		log_info "URL: $preset_url"
+
+		if wget -O "$models_path/$preset_file" "$preset_url"; then
+			log_info "Model installed: $models_path/$preset_file"
+
+			# Create model config YAML
+			cat > "$models_path/$model_name.yaml" << EOF
+name: $model_name
+backend: llama-cpp
+parameters:
+  model: $preset_file
+context_size: $context_size
+threads: $threads
+EOF
+			log_info "Model config created: $models_path/$model_name.yaml"
+		else
+			log_error "Failed to download model"
+			return 1
+		fi
+	elif echo "$model_name" | grep -q "^http"; then
+		# Direct URL download
+		local filename=$(basename "$model_name")
+		log_info "Downloading model from URL..."
+
+		if wget -O "$models_path/$filename" "$model_name"; then
+			log_info "Model installed: $models_path/$filename"
+		else
+			log_error "Failed to download model"
+			return 1
+		fi
+	else
+		log_error "Unknown model or preset: $model_name"
+		# List available presets from UCI
+		local presets=""
+		for section in $(uci show localai 2>/dev/null | grep "=preset" | cut -d. -f2 | cut -d= -f1); do
+			local pname=$(uci_get "$section.name")
+			[ -n "$pname" ] && presets="$presets $pname"
+		done
+		log_info "Available presets:$presets"
+		return 1
+	fi
+}
+
+cmd_model_remove() {
+	load_config
+	require_root
+
+	local model_name="$1"
+	[ -z "$model_name" ] && { echo "Usage: localaictl model-remove <model-name>"; return 1; }
+
+	# Find and remove model files
+	local found=0
+	for ext in gguf bin onnx yaml; do
+		local file="$models_path/$model_name.$ext"
+		if [ -f "$file" ]; then
+			rm -f "$file"
+			log_info "Removed: $file"
+			found=1
+		fi
+	done
+
+	[ "$found" -eq 0 ] && log_warn "Model not found: $model_name"
+}
+
+# =============================================================================
+# COMMANDS
+# =============================================================================
+
+cmd_install() {
+	require_root
+	load_config
+
+	if ! has_lxc; then
+		log_error "LXC not available. Install lxc packages first."
+		exit 1
+	fi
+
+	log_info "Installing LocalAI..."
+
+	# Create directories
+	ensure_dir "$data_path"
+	ensure_dir "$models_path"
+
+	lxc_check_prereqs || exit 1
+	lxc_create_rootfs || exit 1
+
+	uci_set main.enabled '1'
+	/etc/init.d/localai enable
+
+	log_info "LocalAI installed."
+	log_info "Start with: /etc/init.d/localai start"
+	log_info "API endpoint: http://<router-ip>:$api_port/v1"
+	log_info ""
+	log_info "Install a model to get started:"
+	log_info "  localaictl model-install tinyllama   # Lightweight (669MB)"
+	log_info "  localaictl model-install phi2        # Balanced (1.6GB)"
+}
+
+cmd_check() {
+	load_config
+
+	log_info "Checking prerequisites..."
+	if has_lxc; then
+		log_info "LXC: available"
+		lxc_check_prereqs
+	else
+		log_warn "LXC: not available"
+	fi
+
+	# Check memory
+	local mem_total=$(grep MemTotal /proc/meminfo | awk '{print $2}')
+	local mem_gb=$((mem_total / 1024 / 1024))
+	log_info "System memory: ${mem_gb}GB"
+
+	if [ "$mem_gb" -lt 2 ]; then
+		log_warn "Low memory! LocalAI requires at least 2GB RAM for most models"
+	fi
+}
+
+cmd_update() {
+	require_root
+	load_config
+
+	log_info "Updating LocalAI..."
+	lxc_destroy
+	lxc_create_rootfs || exit 1
+
+	if /etc/init.d/localai enabled >/dev/null 2>&1; then
+		/etc/init.d/localai restart
+	else
+		log_info "Update complete. Restart manually to apply."
+	fi
+}
+
+cmd_status() {
+	lxc_status
+}
+
+cmd_logs() {
+	lxc_logs "$@"
+}
+
+cmd_shell() {
+	lxc_shell
+}
+
+cmd_service_run() {
+	require_root
+	load_config
+
+	if ! has_lxc; then
+		log_error "LXC not available"
+		exit 1
+	fi
+
+	lxc_check_prereqs || exit 1
+	lxc_run
+}
+
+cmd_service_stop() {
+	require_root
+	lxc_stop
+}
+
+# Main Entry Point
+case "${1:-}" in
+	install) shift; cmd_install "$@" ;;
+	check) shift; cmd_check "$@" ;;
+	update) shift; cmd_update "$@" ;;
+	status) shift; cmd_status "$@" ;;
+	logs) shift; cmd_logs "$@" ;;
+	shell) shift; cmd_shell "$@" ;;
+	models) shift; cmd_models "$@" ;;
+	model-install) shift; cmd_model_install "$@" ;;
+	model-remove) shift; cmd_model_remove "$@" ;;
+	service-run) shift; cmd_service_run "$@" ;;
+	service-stop) shift; cmd_service_stop "$@" ;;
+	help|--help|-h|'') usage ;;
+	*) echo "Unknown command: $1" >&2; usage >&2; exit 1 ;;
+esac