feat: add nvidia and openrouter platform support

BitFun-Installer/src/data/modelProviders.ts

@@ -25,6 +25,8 @@ export const PROVIDER_DISPLAY_ORDER: string[] = [
   'deepseek',
   'volcengine',
   'siliconflow',
+  'nvidia',
+  'openrouter',
   'minimax',
   'moonshot',
   'anthropic',
@@ -161,6 +163,24 @@ export const PROVIDER_TEMPLATES: Record<string, ProviderTemplate> = {
       },
     ],
   },
+  nvidia: {
+    id: 'nvidia',
+    nameKey: 'model.providers.nvidia.name',
+    descriptionKey: 'model.providers.nvidia.description',
+    baseUrl: 'https://integrate.api.nvidia.com/v1',
+    format: 'openai',
+    models: [],
+    helpUrl: 'https://build.nvidia.com/settings/api-keys',
+  },
+  openrouter: {
+    id: 'openrouter',
+    nameKey: 'model.providers.openrouter.name',
+    descriptionKey: 'model.providers.openrouter.description',
+    baseUrl: 'https://openrouter.ai/api/v1',
+    format: 'openai',
+    models: [],
+    helpUrl: 'https://openrouter.ai/keys',
+  },
 };
 
 export function getOrderedProviders(): ProviderTemplate[] {

BitFun-Installer/src/i18n/locales/en.json

@@ -116,6 +116,14 @@
           "default": "OpenAI Format - Default",
           "anthropic": "Anthropic Format"
         }
+      },
+      "nvidia": {
+        "name": "NVIDIA",
+        "description": "NVIDIA NIM Model Platform"
+      },
+      "openrouter": {
+        "name": "OpenRouter",
+        "description": "OpenRouter Model Platform"
       }
     },
     "modelNameSelectPlaceholder": "Select a model...",

BitFun-Installer/src/i18n/locales/zh.json

@@ -116,6 +116,14 @@
           "default": "OpenAI格式-默认",
           "anthropic": "Anthropic格式"
         }
+      },
+      "nvidia": {
+        "name": "NVIDIA",
+        "description": "NVIDIA NIM 大模型平台"
+      },
+      "openrouter": {
+        "name": "OpenRouter",
+        "description": "OpenRouter 大模型平台"
       }
     },
     "modelNameSelectPlaceholder": "选择模型...",

src/crates/core/src/agentic/image_analysis/types.rs

@@ -120,7 +120,7 @@ impl ImageLimits {
     /// Get limits based on model provider
     pub fn for_provider(provider: &str) -> Self {
         match provider.to_lowercase().as_str() {
-            "openai" | "response" | "responses" => Self {
+            "openai" | "response" | "responses" | "nvidia" | "openrouter" => Self {
                 max_size: 20 * 1024 * 1024, // 20MB
                 max_width: 2048,
                 max_height: 2048,

src/crates/core/src/infrastructure/ai/client.rs

@@ -158,7 +158,10 @@ impl AIClient {
 
     fn build_test_connection_extra_body(&self) -> Option<serde_json::Value> {
         let provider = self.config.format.to_ascii_lowercase();
-        if !matches!(provider.as_str(), "openai" | "response" | "responses") {
+        if !matches!(
+            provider.as_str(),
+            "openai" | "response" | "responses" | "nvidia" | "openrouter"
+        ) {
             return self.config.custom_request_body.clone();
         }
 

src/crates/core/src/util/types/config.rs

@@ -57,7 +57,7 @@ fn resolve_request_url(base_url: &str, provider: &str, model_name: &str) -> Stri
     }
 
     match provider.trim().to_ascii_lowercase().as_str() {
-        "openai" => append_endpoint(&trimmed, "chat/completions"),
+        "openai" | "nvidia" | "openrouter" => append_endpoint(&trimmed, "chat/completions"),
         "response" | "responses" => append_endpoint(&trimmed, "responses"),
         "anthropic" => append_endpoint(&trimmed, "v1/messages"),
         "gemini" | "google" => resolve_gemini_request_url(&trimmed, model_name),
@@ -151,6 +151,22 @@ mod tests {
             "https://api.openbitfun.com/v1beta/models/gemini-2.5-pro:streamGenerateContent?alt=sse"
         );
     }
+
+    #[test]
+    fn resolves_nvidia_request_url() {
+        assert_eq!(
+            resolve_request_url("https://integrate.api.nvidia.com/v1", "nvidia", ""),
+            "https://integrate.api.nvidia.com/v1/chat/completions"
+        );
+    }
+
+    #[test]
+    fn resolves_openrouter_request_url() {
+        assert_eq!(
+            resolve_request_url("https://openrouter.ai/api/v1", "openrouter", ""),
+            "https://openrouter.ai/api/v1/chat/completions"
+        );
+    }
 }
 
 impl TryFrom<AIModelConfig> for AIConfig {

src/web-ui/src/flow_chat/components/modern/VirtualMessageList.tsx

@@ -149,21 +149,16 @@ export const VirtualMessageList = forwardRef<VirtualMessageListRef>((_, ref) =>
     scrollToBottom,
   }), [scrollToTurn, scrollToIndex, scrollToBottom]);
 
-  // ── Core scroll policy: processing → auto-scroll to bottom ────────────
+  // ── Initial scroll to bottom when processing starts ──────────────────
+  // Note: followOutput handles continuous auto-scroll, so we only need
+  // an initial scroll here. The 300ms interval was removed because it
+  // conflicted with followOutput and caused visual jitter.
   useEffect(() => {
     if (!isProcessing) return;
 
     if (virtuosoRef.current) {
-      virtuosoRef.current.scrollTo({ top: 999999999, behavior: 'smooth' });
+      virtuosoRef.current.scrollTo({ top: 999999999, behavior: 'auto' });
     }
-
-    const intervalId = setInterval(() => {
-      if (virtuosoRef.current) {
-        virtuosoRef.current.scrollTo({ top: 999999999, behavior: 'smooth' });
-      }
-    }, 300);
-
-    return () => clearInterval(intervalId);
   }, [isProcessing]);
 
   const handleFollowOutput = useCallback(() => {

src/web-ui/src/infrastructure/config/services/modelConfigs.ts

@@ -181,6 +181,28 @@ export const PROVIDER_TEMPLATES: Record<string, ProviderTemplate> = {
       { url: 'https://api.siliconflow.cn/v1', format: 'openai', note: 'default' },
       { url: 'https://api.siliconflow.cn/v1/messages', format: 'anthropic', note: 'Anthropic' },
     ]
+  },
+
+  nvidia: {
+    id: 'nvidia',
+    name: t('settings/ai-model:providers.nvidia.name'),
+    baseUrl: 'https://integrate.api.nvidia.com/v1',
+    format: 'openai',
+    models: [],
+    requiresApiKey: true,
+    description: t('settings/ai-model:providers.nvidia.description'),
+    helpUrl: 'https://build.nvidia.com/settings/api-keys'
+  },
+
+  openrouter: {
+    id: 'openrouter',
+    name: t('settings/ai-model:providers.openrouter.name'),
+    baseUrl: 'https://openrouter.ai/api/v1',
+    format: 'openai',
+    models: [],
+    requiresApiKey: true,
+    description: t('settings/ai-model:providers.openrouter.description'),
+    helpUrl: 'https://openrouter.ai/keys'
   }
 };
 

src/web-ui/src/infrastructure/config/services/providerCatalog.ts

@@ -63,6 +63,14 @@ export const PROVIDER_URL_CATALOG: ProviderUrlCatalogItem[] = [
       'https://api.siliconflow.cn/v1/messages',
     ],
   },
+  {
+    id: 'nvidia',
+    baseUrl: 'https://integrate.api.nvidia.com/v1',
+  },
+  {
+    id: 'openrouter',
+    baseUrl: 'https://openrouter.ai/api/v1',
+  },
 ];
 
 export function normalizeProviderBaseUrl(url: string): string {

src/web-ui/src/locales/en-US/settings/ai-model.json

@@ -86,6 +86,14 @@
         "default": "OpenAI Format - Default",
         "anthropic": "Anthropic Format"
       }
+    },
+    "nvidia": {
+      "name": "NVIDIA",
+      "description": "NVIDIA NIM Model Platform"
+    },
+    "openrouter": {
+      "name": "OpenRouter",
+      "description": "OpenRouter Model Platform"
     }
   },
   "tabs": {

src/web-ui/src/locales/zh-CN/settings/ai-model.json

@@ -86,6 +86,14 @@
         "default": "OpenAI格式-默认",
         "anthropic": "Anthropic格式"
       }
+    },
+    "nvidia": {
+      "name": "NVIDIA",
+      "description": "NVIDIA NIM 大模型平台"
+    },
+    "openrouter": {
+      "name": "OpenRouter",
+      "description": "OpenRouter 大模型平台"
     }
   },
   "tabs": {