diff --git a/frontend/src/pages/v2/Settings.tsx b/frontend/src/pages/v2/Settings.tsx
index e0124a1..77f8310 100644
--- a/frontend/src/pages/v2/Settings.tsx
+++ b/frontend/src/pages/v2/Settings.tsx
@@ -50,9 +50,9 @@ const LABELS: Record<string, string> = {
   codex_summary_timeout_seconds: "Codex 总结超时（秒）",
   codex_summary_max_retries: "Codex 最大重试次数",
   codex_summary_request_concurrency: "Codex 总结并发数",
-  ai_base_url: "DeepSeek Base URL（备用）",
+  ai_base_url: "DeepSeek Base URL",
   ai_api_key: "DeepSeek API Key",
-  ai_model: "DeepSeek 模型 ID（备用）",
+  ai_model: "DeepSeek 模型 ID",
   ai_timeout_seconds: "DeepSeek 超时（秒）",
   ai_max_retries: "DeepSeek 最大重试次数",
   max_context_chars: "整群直接提交上限（字符）",
@@ -96,14 +96,14 @@ const SETTING_GROUPS = [
   {
     id: "model",
     title: "模型与 Prompt",
-    description: "Codex GPT 主用；单次失败时自动切换到已配置的 DeepSeek 备用。",
+    description: "各群可分别选择聊天分析与海报文案模型。备用 Provider 设为 none 时不自动切换；Codex 思考强度同时用于文字和生图执行。",
     icon: PlugsConnected,
     keys: ["summary_provider_primary", "summary_provider_fallback", "codex_summary_model", "codex_reasoning_effort", "codex_image_model", "codex_summary_timeout_seconds", "codex_summary_max_retries", "ai_base_url", "ai_model", "ai_api_key", "ai_timeout_seconds", "ai_max_retries", "max_context_chars"],
   },
   {
     id: "advanced",
     title: "高级并发设置",
-    description: "控制群任务、微信取数、Codex 总结和 DeepSeek 备用请求的并发上限。",
+    description: "控制群任务、微信取数、Codex 文字处理和 DeepSeek 请求的并发上限。",
     icon: GearSix,
     keys: ["generation_group_concurrency", "wechat_fetch_concurrency", "codex_summary_request_concurrency", "ai_request_concurrency", "wechat_native_stage_timeout_seconds", "wechat_native_submit_timeout_seconds", "wechat_native_poll_interval_seconds"],
   },