diff --git a/康康/AI/AIRuntime.swift b/康康/AI/AIRuntime.swift
index b4f7bba..e9d2153 100644
--- a/康康/AI/AIRuntime.swift
+++ b/康康/AI/AIRuntime.swift
@@ -38,7 +38,7 @@ actor AIRuntime {
     private(set) var mnnStatus: Status = .notReady
     /// MNN 模型目录(下载/旁路导入到 Models/Qwen3.5-2B-MNN)。
     nonisolated static var mnnModelFolder: URL {
-        ModelStore.shared.rootURL.appendingPathComponent("Qwen3.5-2B-MNN", isDirectory: true)
+        ModelStore.shared.localURL(for: .mnnLLM)
     }
 
     // MARK: - 串行推理闸门(§3.1 OOM 防护的真正落地)
@@ -92,8 +92,7 @@ actor AIRuntime {
     func prepare() async throws {
         // 选了 MNN 且模型已就绪才走 MNN;否则(选 MLX,或 MNN 模型尚未下载)回退 MLX,
         // 保证过渡期 App 始终可用。引擎指示器(Phase 5)展示实际生效后端。
-        let mnnReady = FileManager.default.fileExists(
-            atPath: Self.mnnModelFolder.appendingPathComponent("config.json").path)
+        let mnnReady = ModelStore.shared.isComplete(for: .mnnLLM)
         if InferenceEngine.current == .mnn, mnnReady {
             try await prepareMNN()
             return
@@ -147,8 +146,7 @@ actor AIRuntime {
         if mnnStatus == .ready { return }
 
         let folder = Self.mnnModelFolder
-        let config = folder.appendingPathComponent("config.json").path
-        guard FileManager.default.fileExists(atPath: config) else {
+        guard ModelStore.shared.isComplete(for: .mnnLLM) else {
             mnnStatus = .error("MNN 模型未就绪")
             throw AIRuntimeError.notReady
         }
diff --git a/康康/AI/ModelManifest.swift b/康康/AI/ModelManifest.swift
index 18b1e99..951d07e 100644
--- a/康康/AI/ModelManifest.swift
+++ b/康康/AI/ModelManifest.swift
@@ -59,6 +59,19 @@ nonisolated enum ModelManifest {
                 ModelFile(path: "preprocessor_config.json", bytes: 782),
                 ModelFile(path: "video_preprocessor_config.json", bytes: 817),
             ]
+        case .mnnLLM:
+            // taobao-mnn/Qwen3.5-2B-MNN 预转换 MNN 格式(HF API 实测,2026-06)。
+            // 运行时必需:config.json(MNN llm 配置)+ llm_config.json(超参)+ llm.mnn(图)
+            // + llm.mnn.weight(量化权重 ~1.1GB)+ tokenizer.txt + visual.mnn(多模态,文本路径不用但配置含 mllm)。
+            // 排除 README/.gitattributes 与可读 dump(llm.mnn.json / export_args.json)。
+            return [
+                ModelFile(path: "config.json", bytes: 652),
+                ModelFile(path: "llm_config.json", bytes: 8_692),
+                ModelFile(path: "llm.mnn", bytes: 2_148_136),
+                ModelFile(path: "llm.mnn.weight", bytes: 1_176_647_702),
+                ModelFile(path: "tokenizer.txt", bytes: 6_465_727),
+                ModelFile(path: "visual.mnn", bytes: 488_096),
+            ]
         }
     }
 
diff --git a/康康/AI/ModelStore.swift b/康康/AI/ModelStore.swift
index dd31e67..c325ae5 100644
--- a/康康/AI/ModelStore.swift
+++ b/康康/AI/ModelStore.swift
@@ -1,15 +1,19 @@
 import Foundation
 
 nonisolated enum ModelKind: String, CaseIterable {
-    /// 与 HuggingFace mlx-community 仓库名一一对应,也是沙盒 Models/ 下的子目录名。
-    /// 文本 LLM 用 Qwen3.5-2B(多模态权重,走 mlx-swift-lm 的 qwen3_5 → Qwen35Model 文本路径加载)。
-    case llm = "Qwen3.5-2B-4bit"
-    case vl  = "Qwen3-VL-4B-Instruct-4bit"
+    /// 也是沙盒 Models/ 下的子目录名 / CDN 路径段。
+    /// - llm:MLX(GPU)文本兜底,Qwen3.5-2B(多模态权重,走 qwen3_5 文本路径)。
+    /// - vl :MLX(GPU)拍照识别,Qwen3-VL-4B。
+    /// - mnnLLM:MNN(CPU/SME2,挑战赛考核路径)文本,Qwen3.5-2B 预转换 MNN 格式(taobao-mnn)。
+    case llm    = "Qwen3.5-2B-4bit"
+    case vl     = "Qwen3-VL-4B-Instruct-4bit"
+    case mnnLLM = "Qwen3.5-2B-MNN"
 
     var displayName: String {
         switch self {
-        case .llm: return "Qwen3.5-2B"
-        case .vl:  return "Qwen3-VL-4B"
+        case .llm:    return "Qwen3.5-2B (MLX)"
+        case .vl:     return "Qwen3-VL-4B"
+        case .mnnLLM: return "Qwen3.5-2B (MNN/SME2)"
         }
     }
 
diff --git a/康康/Features/Me/ModelManagementView.swift b/康康/Features/Me/ModelManagementView.swift
index 82e1e3f..67f73c8 100644
--- a/康康/Features/Me/ModelManagementView.swift
+++ b/康康/Features/Me/ModelManagementView.swift
@@ -218,8 +218,9 @@ struct ModelManagementView: View {
 
     private func subtitle(_ kind: ModelKind) -> String {
         switch kind {
-        case .llm: return String(appLoc: "文本解读 · 趋势 / 问答")
-        case .vl:  return String(appLoc: "拍照识别报告 → 结构化指标")
+        case .llm:    return String(appLoc: "文本解读 · 趋势 / 问答(MLX 兜底)")
+        case .vl:     return String(appLoc: "拍照识别报告 → 结构化指标")
+        case .mnnLLM: return String(appLoc: "文本解读 · MNN + SME2 端侧加速")
         }
     }
 
diff --git a/康康Tests/ModelManifestTests.swift b/康康Tests/ModelManifestTests.swift
index acbab04..eaecdaf 100644
--- a/康康Tests/ModelManifestTests.swift
+++ b/康康Tests/ModelManifestTests.swift
@@ -20,6 +20,28 @@ struct ModelManifestTests {
         #expect(ModelManifest.totalBytes(for: .vl) == 3_109_729_929)
     }
 
+    @Test func mnnHasSixFunctionalFiles() {
+        #expect(ModelManifest.files(for: .mnnLLM).count == 6)
+    }
+
+    @Test func mnnTotalBytesMatchesManifest() {
+        #expect(ModelManifest.totalBytes(for: .mnnLLM) == 1_185_759_005)
+    }
+
+    @Test func mnnHasEssentialRuntimeFiles() {
+        let names = ModelManifest.files(for: .mnnLLM).map(\.path)
+        #expect(names.contains("config.json"))
+        #expect(names.contains("llm.mnn"))
+        #expect(names.contains("llm.mnn.weight"))
+        #expect(names.contains("tokenizer.txt"))
+    }
+
+    @Test func mnnFileURLUsesRepoPath() {
+        let file = ModelFile(path: "config.json", bytes: 652)
+        let url = ModelManifest.fileURL(for: .mnnLLM, file: file)
+        #expect(url.absoluteString == "https://file.myv0.com/Qwen3.5-2B-MNN/config.json")
+    }
+
     @Test func excludesReadmeAndGitattributes() {
         for kind in [ModelKind.llm, .vl] {
             let names = ModelManifest.files(for: kind).map(\.path)