feat(web): expose cheap_model and smart_routing_cascade in settings UI

src/channels/web/static/app.js

@@ -4705,6 +4705,8 @@ var INFERENCE_SETTINGS = [
       { key: 'llm_backend', label: 'cfg.llm_backend.label', description: 'cfg.llm_backend.desc',
         type: 'select', options: ['nearai', 'anthropic', 'openai', 'ollama', 'openai_compatible', 'tinfoil', 'bedrock'] },
       { key: 'selected_model', label: 'cfg.selected_model.label', description: 'cfg.selected_model.desc', type: 'text' },
+      { key: 'cheap_model', label: 'cfg.cheap_model.label', description: 'cfg.cheap_model.desc', type: 'text' },
+      { key: 'smart_routing_cascade', label: 'cfg.smart_routing_cascade.label', description: 'cfg.smart_routing_cascade.desc', type: 'boolean' },
       { key: 'ollama_base_url', label: 'cfg.ollama_base_url.label', description: 'cfg.ollama_base_url.desc', type: 'text',
         showWhen: { key: 'llm_backend', value: 'ollama' } },
       { key: 'openai_compatible_base_url', label: 'cfg.openai_compatible_base_url.label', description: 'cfg.openai_compatible_base_url.desc', type: 'text',
@@ -5097,7 +5099,7 @@ function renderStructuredSettingsRow(def, value, activeValue) {
   return row;
 }
 
-var RESTART_REQUIRED_KEYS = ['llm_backend', 'selected_model', 'ollama_base_url', 'openai_compatible_base_url',
+var RESTART_REQUIRED_KEYS = ['llm_backend', 'selected_model', 'cheap_model', 'smart_routing_cascade', 'ollama_base_url', 'openai_compatible_base_url',
   'bedrock_region', 'bedrock_cross_region', 'bedrock_profile', 'embeddings.enabled', 'embeddings.provider', 'embeddings.model',
   'agent.auto_approve_tools', 'tunnel.provider', 'tunnel.public_url', 'gateway.rate_limit', 'gateway.max_connections'];
 

src/channels/web/static/i18n/en.js

@@ -403,6 +403,10 @@ I18n.register('en', {
   'cfg.llm_backend.desc': 'LLM inference provider',
   'cfg.selected_model.label': 'Model',
   'cfg.selected_model.desc': 'Model name or ID for the selected backend',
+  'cfg.cheap_model.label': 'Cheap Model',
+  'cfg.cheap_model.desc': 'Cheap/fast model for smart routing (lightweight tasks)',
+  'cfg.smart_routing_cascade.label': 'Smart Routing Cascade',
+  'cfg.smart_routing_cascade.desc': 'Retry with primary model if cheap model response seems uncertain',
   'cfg.ollama_base_url.label': 'Ollama URL',
   'cfg.ollama_base_url.desc': 'Base URL for Ollama API',
   'cfg.openai_compatible_base_url.label': 'OpenAI-compatible URL',

src/channels/web/static/i18n/zh-CN.js

@@ -402,6 +402,10 @@ I18n.register('zh-CN', {
   'cfg.llm_backend.desc': 'LLM 推理提供商',
   'cfg.selected_model.label': '模型',
   'cfg.selected_model.desc': '所选后端的模型名称或 ID',
+  'cfg.cheap_model.label': '廉价模型',
+  'cfg.cheap_model.desc': '用于智能路由的廉价/快速模型（轻量级任务）',
+  'cfg.smart_routing_cascade.label': '智能路由级联',
+  'cfg.smart_routing_cascade.desc': '当廉价模型回答不确定时，使用主模型重试',
   'cfg.ollama_base_url.label': 'Ollama URL',
   'cfg.ollama_base_url.desc': 'Ollama API 基础 URL',
   'cfg.openai_compatible_base_url.label': 'OpenAI 兼容 URL',

src/config/llm.rs

@@ -213,13 +213,22 @@ impl LlmConfig {
 
         let request_timeout_secs = parse_optional_env("LLM_REQUEST_TIMEOUT_SECS", 120)?;
 
-        // Generic cheap model (works with any backend).
+        // Generic cheap model: env var > settings > None.
         // Falls back to NearAI-specific cheap_model in provider chain logic.
-        let cheap_model = optional_env("LLM_CHEAP_MODEL")?;
-
-        // Generic smart routing cascade flag.
-        // Defaults to true. Overrides NearAI-specific smart_routing_cascade.
-        let smart_routing_cascade = parse_optional_env("SMART_ROUTING_CASCADE", true)?;
+        let cheap_model = optional_env("LLM_CHEAP_MODEL")?
+            .or_else(|| settings.cheap_model.clone());
+
+        // Generic smart routing cascade flag: env var > settings > true.
+        // Overrides NearAI-specific smart_routing_cascade.
+        let smart_routing_cascade = optional_env("SMART_ROUTING_CASCADE")?
+            .map(|v| v.parse::<bool>())
+            .transpose()
+            .map_err(|_| ConfigError::InvalidValue {
+                key: "SMART_ROUTING_CASCADE".into(),
+                message: "expected true or false".into(),
+            })?
+            .or(settings.smart_routing_cascade)
+            .unwrap_or(true);
 
         Ok(Self {
             backend: if is_nearai {

src/settings.rs

@@ -84,6 +84,15 @@ pub struct Settings {
     #[serde(default)]
     pub selected_model: Option<String>,
 
+    /// Cheap/fast model for smart routing (lightweight tasks like heartbeat, routing).
+    #[serde(default)]
+    pub cheap_model: Option<String>,
+
+    /// Enable cascade mode for smart routing (retry with primary if cheap model
+    /// response seems uncertain). When None, defaults to true.
+    #[serde(default)]
+    pub smart_routing_cascade: Option<bool>,
+
     // === Step 5: Embeddings ===
     /// Embeddings configuration.
     #[serde(default)]