jinzd-ai-cli 0.4.328 → 0.4.330

package/README.md

@@ -134,10 +134,10 @@ aicli user delete x          # Delete user
 | **DeepSeek** | deepseek-flash (default, multimodal), deepseek-v4-pro | [platform.deepseek.com](https://platform.deepseek.com) |
 | **OpenAI** | GPT-5.4, GPT-5, GPT-4.1, o3, o4-mini | [platform.openai.com](https://platform.openai.com) |
 | **OpenRouter** | 300+ models (Claude, GPT, Gemini, Llama, Qwen, Mistral...) | [openrouter.ai](https://openrouter.ai) |
-| **Zhipu** | GLM-4, GLM-5 | [open.bigmodel.cn](https://open.bigmodel.cn) |
-| **Kimi** | Kimi K2.6, K2.5, K2 Thinking | [platform.moonshot.cn](https://platform.moonshot.cn) |
-| **Qwen** | qwen-plus, qwen-max, qwen-turbo, qwen-coder-plus | [bailian.console.aliyun.com](https://bailian.console.aliyun.com) |
-| **MiniMax** | MiniMax-M3 (default), M2.7, M2.5, M2.1, M2 | [platform.minimaxi.com](https://platform.minimaxi.com) |
+| **Zhipu** | glm-5.3-flash (default), glm-5.3-flashx, glm-5.3, glm-4.7-flash (free) | [open.bigmodel.cn](https://open.bigmodel.cn) |
+| **Kimi** | kimi-k2.7-code (default), kimi-k2.7-code-highspeed, kimi-k3 | [platform.moonshot.cn](https://platform.moonshot.cn) |
+| **Qwen** | qwen3.8-flash (default), qwen3.7-plus, qwen3.8-max, qwen-flash | [bailian.console.aliyun.com](https://bailian.console.aliyun.com) |
+| **MiniMax** | MiniMax-M3 (default) | [platform.minimaxi.com](https://platform.minimaxi.com) |
 | **Ollama** | Any locally installed model (Llama, Qwen, Gemma, Mistral...) | No API key — [ollama.com](https://ollama.com) |
 
 Any OpenAI-compatible API can also be used via `customBaseUrls` in config.

package/README.zh-CN.md

@@ -122,10 +122,10 @@ aicli user delete x          # 删除用户
 | **DeepSeek** | deepseek-flash（默认，多模态）, deepseek-v4-pro | [platform.deepseek.com](https://platform.deepseek.com) |
 | **OpenAI** | GPT-5.4, GPT-5, GPT-4.1, o3, o4-mini | [platform.openai.com](https://platform.openai.com) |
 | **OpenRouter** | 300+ 模型（Claude/GPT/Gemini/Llama/Qwen/Mistral...） | [openrouter.ai](https://openrouter.ai) |
-| **Zhipu（智谱）** | GLM-4, GLM-5 | [open.bigmodel.cn](https://open.bigmodel.cn) |
-| **Kimi** | Kimi K2.6, K2.5, K2 Thinking | [platform.moonshot.cn](https://platform.moonshot.cn) |
-| **Qwen** | qwen-plus, qwen-max, qwen-turbo, qwen-coder-plus | [bailian.console.aliyun.com](https://bailian.console.aliyun.com) |
-| **MiniMax** | MiniMax-M3（默认）、M2.7、M2.5、M2.1、M2 | [platform.minimaxi.com](https://platform.minimaxi.com) |
+| **Zhipu（智谱）** | glm-5.3-flash（默认）、glm-5.3-flashx、glm-5.3、glm-4.7-flash（免费） | [open.bigmodel.cn](https://open.bigmodel.cn) |
+| **Kimi** | kimi-k2.7-code（默认）、kimi-k2.7-code-highspeed、kimi-k3 | [platform.moonshot.cn](https://platform.moonshot.cn) |
+| **Qwen** | qwen3.8-flash（默认）、qwen3.7-plus、qwen3.8-max、qwen-flash | [bailian.console.aliyun.com](https://bailian.console.aliyun.com) |
+| **MiniMax** | MiniMax-M3（默认） | [platform.minimaxi.com](https://platform.minimaxi.com) |
 | **Ollama** | 任意本地模型（Llama、Qwen、Gemma、Mistral...） | 无需 API Key — [ollama.com](https://ollama.com) |
 
 也可通过 `customBaseUrls` 配置接入任意 OpenAI 兼容 API。

package/dist/{batch-GE6XWXPK.js → batch-GXITPLER.js}

@@ -4,12 +4,12 @@ import {
 } from "./chunk-RS7SKYTS.js";
 import {
   ConfigManager
-} from "./chunk-OVEZ6HB5.js";
+} from "./chunk-QU74ONFF.js";
 import "./chunk-TU3L3PW7.js";
 import {
   atomicWriteFileSync
 } from "./chunk-BMLQPSHP.js";
-import "./chunk-NOQKCHIQ.js";
+import "./chunk-AJITHMRE.js";
 
 // src/cli/batch.ts
 import Anthropic from "@anthropic-ai/sdk";

package/dist/{chunk-IEKVRC2E.js → chunk-265M6ZLH.js}

@@ -8,13 +8,13 @@ import {
   listHooks,
   listInstalledPlugins,
   resetStats
-} from "./chunk-B4EBCSKM.js";
+} from "./chunk-HMQA2R3O.js";
 import {
   ProviderRegistry
-} from "./chunk-NKTCITLH.js";
+} from "./chunk-GEC3MEVQ.js";
 import {
   ConfigManager
-} from "./chunk-OVEZ6HB5.js";
+} from "./chunk-QU74ONFF.js";
 import {
   getGitRoot
 } from "./chunk-HOSJZMQS.js";
@@ -26,7 +26,7 @@ import {
   MCP_PROJECT_CONFIG_NAME,
   MEMORY_FILE_NAME,
   VERSION
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/diagnostics/doctor-report.ts
 import { existsSync as existsSync3, statSync as statSync2 } from "fs";

package/dist/{chunk-BRZZ3QAI.js → chunk-2BAFLDDL.js}

@@ -16,7 +16,6 @@ function deepSeekCnyPricing(cacheRead, input, output) {
     output: output / CNY_PER_USD
   };
 }
-var DEEPSEEK_V4_PRO_ROUTED_TO_FLASH_AT = /* @__PURE__ */ new Date("2026-09-14T04:00:00.000Z");
 var DEEPSEEK_V41_FLASH_ROUTING_CONFIRMED_AT = /* @__PURE__ */ new Date("2026-09-11T00:00:00.000Z");
 function isDeepSeekPeakHour(at) {
   const weekday = at.getUTCDay();
@@ -32,7 +31,6 @@ function getDeepSeekV4Pricing(model, at) {
   const isPro = model.startsWith("deepseek-v4-pro");
   const isFlash = model.startsWith("deepseek-v4-flash");
   if (!isPro && !isFlash) return null;
-  if (isPro && at >= DEEPSEEK_V4_PRO_ROUTED_TO_FLASH_AT) return deepSeekFlash41Pricing(at);
   if (isFlash && at >= DEEPSEEK_V41_FLASH_ROUTING_CONFIRMED_AT) return deepSeekFlash41Pricing(at);
   if (at < DEEPSEEK_TIME_PRICING_EFFECTIVE_AT) {
     return isPro ? deepSeekCnyPricing(0.025, 3, 6) : deepSeekCnyPricing(0.02, 1, 2);
@@ -111,6 +109,12 @@ var PRICING_TABLE = {
   // Kimi K3（2026-07 旗舰，1M 上下文）官方定价 platform.kimi.ai/pricing/chat-k3：
   // 输入 cache-miss $3.00 / cache-hit $0.30，输出 $15.00（/MTok）。
   "kimi-k3": { input: 3, output: 15, cacheRead: 0.3 },
+  // K2.6 / K2.7 Code：官方 platform.kimi.ai/docs/pricing/chat（2026-10-18 核对，USD/MTok）。
+  // 国内站人民币价 ¥6.5/¥27（缓存 ¥1.1 / ¥1.3）按 ¥6.9 折算与之相差 <3%，沿用 K3 那行的美元口径。
+  // 此前 K2.6 没有单列，前缀落到 'kimi-k2'（$0.6/$2.5），输出低报约 37%。
+  "kimi-k2.6": { input: 0.95, output: 4, cacheRead: 0.16 },
+  "kimi-k2.7-code": { input: 0.95, output: 4, cacheRead: 0.19 },
+  "kimi-k2.7-code-highspeed": { input: 1.9, output: 8, cacheRead: 0.38 },
   // ── Zhipu GLM ─────────────────────────────────────────────────
   "glm-4-plus": { input: 0.7, output: 0.7 },
   "glm-4": { input: 0.14, output: 0.14 },
@@ -132,6 +136,11 @@ var PRICING_TABLE = {
   // 没公布截止日期，静态表表达不了这个时间维度；高估比低估安全——预算告警早触发
   // 好过晚触发。折扣期内 `/cost` 会比实际偏高约一倍，这是刻意的取舍。
   "glm-5.3-flash": { input: 0.116, output: 0.406, cacheRead: 0.033 },
+  // GLM-5.3-FlashX：官方定价页（docs.bigmodel.cn/cn/guide/start/pricing，2026-10-18 浏览器读取）
+  // 输入 2 元、输出 7 元、缓存命中 0.57 元。必须单列——否则前缀匹配会落到上面的 glm-5.3-flash，低报 2.5 倍。
+  "glm-5.3-flashx": { input: 2 / CNY_PER_USD, output: 7 / CNY_PER_USD, cacheRead: 0.57 / CNY_PER_USD },
+  // GLM-4.7-Flash：官方免费（输入/输出/缓存全免）。必须单列——否则前缀落到 'glm-4'（$0.14）。
+  "glm-4.7-flash": { input: 0, output: 0, cacheRead: 0 },
   // GLM-5.2：与 GLM-5.3 同价（同一定价页：8 / 28 / 缓存命中 2 元），口径修正同上。
   "glm-5.2": { input: 8 / CNY_PER_USD, output: 28 / CNY_PER_USD, cacheRead: 2 / CNY_PER_USD },
   // GLM-5V-Turbo（2026-04 视觉编程旗舰）官方直连 API 定价 $1.20 / $4.00。
@@ -141,8 +150,26 @@ var PRICING_TABLE = {
   "glm-z1-air": { input: 0.2, output: 0.6 },
   "glm-z1-flash": { input: 0, output: 0 },
   // ── MiniMax (海螺) ────────────────────────────────────────────
-  // GAP (v0.4.184): MiniMax-M3 / M2.x are unpriced — no verified rate anchor yet.
-  // Unpriced → `aicli usage` shows "—" (unknown), not a false $0.00. Add when confirmed.
+  // 官方定价页 platform.minimax.cn/docs/guides/pricing（2026-10-18 核对，元/百万 tokens），按 ¥6.9 折算。
+  // 本 provider 打的是国内站 api.minimaxi.com，按人民币计费。
+  // M3 记的是「永久五折」后的价（官方原文标注永久，不同于智谱的限时折扣），≤512K 输入档；
+  // >512K 那档见 computeCost 的 LONG_PROMPT_TIERS。M3 未公布缓存写入价，缺省按输入价（偏高估）。
+  "minimax-m3": { input: 2.1 / CNY_PER_USD, output: 8.4 / CNY_PER_USD, cacheRead: 0.42 / CNY_PER_USD },
+  "minimax-m2.7": { input: 2.1 / CNY_PER_USD, output: 8.4 / CNY_PER_USD, cacheRead: 0.42 / CNY_PER_USD, cacheWrite: 2.625 / CNY_PER_USD },
+  "minimax-m2.7-highspeed": { input: 4.2 / CNY_PER_USD, output: 16.8 / CNY_PER_USD, cacheRead: 0.42 / CNY_PER_USD, cacheWrite: 2.625 / CNY_PER_USD },
+  // M2.5 / M2.1 / M2（历史模型）同价；highspeed 变体输入输出翻倍、缓存同价。
+  "minimax-m2": { input: 2.1 / CNY_PER_USD, output: 8.4 / CNY_PER_USD, cacheRead: 0.21 / CNY_PER_USD, cacheWrite: 2.625 / CNY_PER_USD },
+  "minimax-m2.5-highspeed": { input: 4.2 / CNY_PER_USD, output: 16.8 / CNY_PER_USD, cacheRead: 0.21 / CNY_PER_USD, cacheWrite: 2.625 / CNY_PER_USD },
+  "minimax-m2.1-highspeed": { input: 4.2 / CNY_PER_USD, output: 16.8 / CNY_PER_USD, cacheRead: 0.21 / CNY_PER_USD, cacheWrite: 2.625 / CNY_PER_USD },
+  // ── Qwen（阿里云百炼，北京地域）───────────────────────────────────
+  // 官方 help.aliyun.com/zh/model-studio/model-pricing（2026-10-18 核对，元/百万 tokens，原价），按 ¥6.9 折算。
+  // 思考与非思考输出同价。qwen3.7-plus 当前限时 8 折，记原价（高估方向，同智谱 glm-5.3-flash 的取舍）。
+  // 官方未给出各模型的缓存单价，不填 cacheRead，命中按输入全价算（高估方向）。
+  // 长提示分档（qwen3.7-plus >256K、qwen-flash >128K / >256K）见 LONG_PROMPT_TIERS；此处为最低档。
+  "qwen3.8-max": { input: 12 / CNY_PER_USD, output: 36 / CNY_PER_USD },
+  "qwen3.7-plus": { input: 2 / CNY_PER_USD, output: 8 / CNY_PER_USD },
+  "qwen3.8-flash": { input: 0.8 / CNY_PER_USD, output: 2.7 / CNY_PER_USD },
+  "qwen-flash": { input: 0.15 / CNY_PER_USD, output: 1.5 / CNY_PER_USD },
   // ── OpenRouter (pass-through — actual cost depends on underlying model) ──
   // 绝大多数 id 仍**刻意留空**：真实价由底层模型 + OpenRouter 加价决定，静态表覆盖不了
   // 396 个模型，未收录的照旧显示 "—"（未知）而不是假的 $0.00。
@@ -197,11 +224,21 @@ function getPricing(provider, model, at = /* @__PURE__ */ new Date()) {
   return null;
 }
 var CLAUDE_HAIKU_55_LONG = { input: 0.5, output: 2.5, cacheWrite: 0.625, cacheRead: 0.05 };
-var CLAUDE_HAIKU_55_TIER_TOKENS = 1e5;
+var LONG_PROMPT_TIERS = [
+  { prefix: "claude-haiku-5-5", overTokens: 1e5, pricing: CLAUDE_HAIKU_55_LONG },
+  {
+    prefix: "minimax-m3",
+    overTokens: 512e3,
+    pricing: { input: 4.2 / CNY_PER_USD, output: 16.8 / CNY_PER_USD, cacheRead: 0.84 / CNY_PER_USD }
+  },
+  { prefix: "qwen3.7-plus", overTokens: 256e3, pricing: { input: 6 / CNY_PER_USD, output: 24 / CNY_PER_USD } },
+  { prefix: "qwen-flash", overTokens: 256e3, pricing: { input: 1.2 / CNY_PER_USD, output: 12 / CNY_PER_USD } },
+  { prefix: "qwen-flash", overTokens: 128e3, pricing: { input: 0.6 / CNY_PER_USD, output: 6 / CNY_PER_USD } }
+];
 function longPromptTier(model, usage) {
-  if (!model.toLowerCase().startsWith("claude-haiku-5-5")) return null;
+  const key = model.toLowerCase();
   const prompt = usage.inputTokens + (usage.cacheCreationTokens ?? 0) + (usage.cacheReadTokens ?? 0);
-  return prompt > CLAUDE_HAIKU_55_TIER_TOKENS ? CLAUDE_HAIKU_55_LONG : null;
+  return LONG_PROMPT_TIERS.find((t) => key.startsWith(t.prefix) && prompt > t.overTokens)?.pricing ?? null;
 }
 function computeCost(provider, model, usage, at = /* @__PURE__ */ new Date()) {
   const p = longPromptTier(model, usage) ?? getPricing(provider, model, at);

package/dist/{chunk-JXGE4T4U.js → chunk-2SYDIWRU.js}

@@ -12,7 +12,7 @@ import {
   MEMORY_FILE_NAME,
   MEMORY_MAX_CHARS,
   MEMORY_STORE_FILE_NAME
-} from "./chunk-LOS7T2JS.js";
+} from "./chunk-3YVZJHTF.js";
 
 // src/memory/persistent-memory.ts
 import { copyFileSync, existsSync, mkdirSync, readFileSync, statSync } from "fs";

package/dist/{chunk-LOS7T2JS.js → chunk-3YVZJHTF.js}

@@ -1,5 +1,5 @@
 // src/core/constants.ts
-var VERSION = "0.4.328";
+var VERSION = "0.4.330";
 var APP_NAME = "ai-cli";
 var CONFIG_DIR_NAME = ".aicli";
 var CONFIG_FILE_NAME = "config.json";

package/dist/{chunk-NOQKCHIQ.js → chunk-AJITHMRE.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 
 // src/core/constants.ts
-var VERSION = "0.4.328";
+var VERSION = "0.4.330";
 var APP_NAME = "ai-cli";
 var CONFIG_DIR_NAME = ".aicli";
 var CONFIG_FILE_NAME = "config.json";

package/dist/{chunk-T4K4RZKR.js → chunk-CCFCPAKN.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   VERSION
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/cli/ci-format.ts
 var DEFAULT_CI_THRESHOLDS = {

package/dist/{chunk-IOLDD5ZA.js → chunk-DU44SD6A.js}

@@ -4,7 +4,7 @@ import {
 } from "./chunk-QMFMST2T.js";
 import {
   TEST_TIMEOUT
-} from "./chunk-LOS7T2JS.js";
+} from "./chunk-3YVZJHTF.js";
 
 // src/tools/builtin/run-tests.ts
 import { spawn as spawn2 } from "child_process";

package/dist/{chunk-E3RJEQZA.js → chunk-DX2U6QK4.js}

@@ -13,7 +13,7 @@ import {
   MEMORY_FILE_NAME,
   MEMORY_MAX_CHARS,
   MEMORY_STORE_FILE_NAME
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/memory/persistent-memory.ts
 import { copyFileSync, existsSync, mkdirSync, readFileSync, statSync } from "fs";

package/dist/{chunk-NKTCITLH.js → chunk-GEC3MEVQ.js}

@@ -15,7 +15,7 @@ import {
 } from "./chunk-BMLQPSHP.js";
 import {
   DEFAULT_MAX_TOKENS
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/providers/claude.ts
 import Anthropic from "@anthropic-ai/sdk";
@@ -1743,7 +1743,9 @@ var DeepSeekProvider = class extends OpenAICompatibleProvider {
       },
       {
         id: "deepseek-v4-pro",
-        displayName: "DeepSeek V4 Pro (text-only; routes to V4.1 Flash from 2026-09-14)",
+        // 2026-09-11 官方撤回了「9-14 起路由到 Flash」，Pro 照常服务、按 Pro 价计费（见 pricing.ts）。
+        // 价格约为 Flash 的 4 倍且不收图片，但 Flash 替代不了它的能力档，故保留。
+        displayName: "DeepSeek V4 Pro (0813, text-only)",
         contextWindow: 1048576,
         supportsStreaming: true,
         supportsThinking: true
@@ -1802,20 +1804,24 @@ var DeepSeekProvider = class extends OpenAICompatibleProvider {
 var ZHIPU_CONFIG = {
   id: "zhipu",
   displayName: "Zhipu (GLM)",
-  defaultModel: "glm-4.6",
+  defaultModel: "glm-5.3-flash",
   apiKeyEnvVar: "AICLI_API_KEY_ZHIPU",
   baseUrl: "https://open.bigmodel.cn/api/paas/v4",
   defaultTimeout: 3e5,
   models: [
     { id: "glm-5.3", displayName: "GLM-5.3 (2026-08 Flagship, 1M, Coding/Security, \u59CB\u7EC8\u601D\u8003)", contextWindow: 1048576, supportsStreaming: true, supportsThinking: true },
+    // GLM-5.3-FlashX（2026-10）：Flash 与 5.3 之间的一档，¥2/¥7、缓存命中 ¥0.57（官方定价页）。
+    // 2026-10-18 实测：同样始终思考（`disabled` → 400/1210），`low` 档 200，base64 图片可识别。
+    { id: "glm-5.3-flashx", displayName: "GLM-5.3 FlashX (\u539F\u751F\u591A\u6A21\u6001, 1M, \u4EF7\u683C\u7EA6\u4E3A 5.3 \u7684 1/4, \u59CB\u7EC8\u601D\u8003)", contextWindow: 1048576, supportsStreaming: true, supportsThinking: true },
     { id: "glm-5.3-flash", displayName: "GLM-5.3 Flash (\u539F\u751F\u591A\u6A21\u6001, 1M, \u8F93\u5165\u7EA6\u4E3A 5.3 \u7684 1/10, \u59CB\u7EC8\u601D\u8003)", contextWindow: 1048576, supportsStreaming: true, supportsThinking: true },
-    { id: "glm-5.2", displayName: "GLM-5.2 (2026-06 Flagship, 1M, Coding)", contextWindow: 1048576, supportsStreaming: true, supportsThinking: true },
-    { id: "glm-5v-turbo", displayName: "GLM-5V-Turbo (Multimodal Vision + Coding)", contextWindow: 202752, supportsStreaming: true },
-    // 2026-10-07 清理：移除 glm-5.1-reasoning / glm-5.1-air（官方模型总览与发布记录里**从未出现过**，
-    // 当初取自第三方标注，同 v0.4.198 的 `glm-5.2[1m]`）与 glm-5 / glm-5.1（已被 glm-5.2 / 5.3 取代，
-    // 要便宜有 glm-5.3-flash；glm-5 这里标的 128K 与官方 200K 也不符）。后两者端点仍可用，`/model` 照收，
-    // pricing.ts 保留其费率供历史成本回算。
-    { id: "glm-4.6", displayName: "GLM-4.6 (200K, \u4E2D\u6587\u5199\u4F5C\u63A8\u8350)", contextWindow: 204800, supportsStreaming: true }
+    // 免费（输入/输出/缓存全免，官方定价页「更多文本模型」）。2026-10-18 实测 200，但慢（一句 ok 约 15s）。
+    { id: "glm-4.7-flash", displayName: "GLM-4.7 Flash (\u514D\u8D39, 200K)", contextWindow: 2e5, supportsStreaming: true }
+    // 2026-10-18 按「老模型没有价格优势就下架」清理（端点仍可用，`/model` 照收，pricing.ts 保留费率供历史回算）：
+    //   - glm-5.2：与 glm-5.3 同价（¥8/¥28/缓存 ¥2），已被取代
+    //   - glm-5v-turbo：¥5/¥22 起，比同样原生多模态的 glm-5.3-flash（¥0.8/¥2.8）贵 6–8 倍
+    //   - glm-4.6：已不在官方定价页上，且旧价高于 glm-5.3-flash
+    // 更早的 2026-10-07 清理：glm-5.1-reasoning / glm-5.1-air 官方从未存在（取自第三方标注，同 v0.4.198
+    // 的 `glm-5.2[1m]`），glm-5 / glm-5.1 已被取代。
   ]
 };
 var REASONING_EFFORT_MODELS = /^glm-5/i;
@@ -1881,7 +1887,10 @@ var KimiProvider = class _KimiProvider extends OpenAICompatibleProvider {
    */
   static TEMP_1_MODELS = /* @__PURE__ */ new Set([
     "kimi-k3",
+    "kimi-k2.7-code",
+    "kimi-k2.7-code-highspeed",
     "kimi-k2.6",
+    // 已 404 的旧 code 仍留在这里：config 自愈迁移之外的显式 `/model kimi-k2.5` 不该先撞 temperature 400
     "kimi-k2.5",
     "kimi-k2-thinking",
     "kimi-k2-thinking-turbo"
@@ -1902,7 +1911,7 @@ var KimiProvider = class _KimiProvider extends OpenAICompatibleProvider {
   info = {
     id: "kimi",
     displayName: "Kimi (Moonshot AI)",
-    defaultModel: "kimi-k2.6",
+    defaultModel: "kimi-k2.7-code",
     apiKeyEnvVar: "AICLI_API_KEY_KIMI",
     requiresApiKey: true,
     baseUrl: this.defaultBaseUrl,
@@ -1912,39 +1921,33 @@ var KimiProvider = class _KimiProvider extends OpenAICompatibleProvider {
       //   - contextWindow 1M（1048576）；code 就是 `kimi-k3`（UI 名 "K3 · Max" ≠ API code）
       //   - temperature 固定 1.0（已并入 TEMP_1_MODELS）
       //   - vision 走 base64 / ms://file-id（不收公网 URL）——本版未接图片输入路径，仅文本/工具
-      // 默认仍保持 kimi-k2.6（K3 输出 $15/MTok 偏贵，按需 /model kimi-k3）。
+      // 默认用 kimi-k2.7-code（K3 输出 $15/MTok，约为它的 4 倍，按需 /model kimi-k3）。
       {
         id: "kimi-k3",
         displayName: "Kimi K3",
         contextWindow: 1048576,
         supportsStreaming: true
       },
-      // Kimi K2.6 / K2.5 系列（最新一代，推荐）
+      // Kimi K2.7 Code（官方 platform.kimi.com/docs/guide/kimi-k2-7-code-quickstart，2026-10-18 核对）：
+      //   - 256K 上下文；虽名为 Coding 模型，官方写明也支持对话与 Agent 任务
+      //   - 思考默认开启且**关不掉**（关闭报错）——本 provider 不发 thinking 字段，正好不踩
+      //   - temperature 固定 1.0（已并入 TEMP_1_MODELS）；tool_choice 只收 auto / none
+      //   - 价格与 K2.6 相同（$0.95/$4.00），各项基准提升 10–30%，故取代 K2.6 成为默认
       {
-        id: "kimi-k2.6",
-        displayName: "Kimi K2.6",
+        id: "kimi-k2.7-code",
+        displayName: "Kimi K2.7 Code",
         contextWindow: 262144,
         supportsStreaming: true
       },
+      // 同一模型的高速版：输出约 5–6 倍速，价格翻倍。
       {
-        id: "kimi-k2.5",
-        displayName: "Kimi K2.5",
-        contextWindow: 262144,
-        supportsStreaming: true
-      },
-      // Kimi K2 Thinking 系列（支持多步推理 + 工具调用）
-      {
-        id: "kimi-k2-thinking",
-        displayName: "Kimi K2 Thinking",
-        contextWindow: 262144,
-        supportsStreaming: true
-      },
-      {
-        id: "kimi-k2-thinking-turbo",
-        displayName: "Kimi K2 Thinking Turbo",
+        id: "kimi-k2.7-code-highspeed",
+        displayName: "Kimi K2.7 Code Highspeed (2x price, ~5x speed)",
         contextWindow: 262144,
         supportsStreaming: true
       }
+      // 2026-10-18 下架：kimi-k2.6 与 K2.7 Code 同价、已被取代（端点仍可用，`/model` 照收）；
+      // kimi-k2.5 / kimi-k2-thinking / kimi-k2-thinking-turbo 端点已 404，由 config 自愈迁移到 kimi-k2.7-code。
     ]
   };
   /**
@@ -2056,11 +2059,10 @@ var MINIMAX_CONFIG = {
   apiKeyEnvVar: "AICLI_API_KEY_MINIMAX",
   baseUrl: "https://api.minimaxi.com/v1",
   models: [
-    { id: "MiniMax-M3", displayName: "MiniMax M3\uFF08\u65D7\u8230\uFF0C1M \u4E0A\u4E0B\u6587\uFF09", contextWindow: 1e6, supportsStreaming: true },
-    { id: "MiniMax-M2.7", displayName: "MiniMax M2.7", contextWindow: 1e6, supportsStreaming: true },
-    { id: "MiniMax-M2.7-highspeed", displayName: "MiniMax M2.7 Highspeed", contextWindow: 1e6, supportsStreaming: true },
-    { id: "MiniMax-M2.5", displayName: "MiniMax M2.5", contextWindow: 1e6, supportsStreaming: true },
-    { id: "MiniMax-M2.5-highspeed", displayName: "MiniMax M2.5 Highspeed", contextWindow: 1e6, supportsStreaming: true }
+    { id: "MiniMax-M3", displayName: "MiniMax M3\uFF08\u65D7\u8230\uFF0C1M \u4E0A\u4E0B\u6587\uFF09", contextWindow: 1e6, supportsStreaming: true }
+    // 2026-10-18 按「老模型没有价格优势就下架」清理（官方定价页 platform.minimax.cn/docs/guides/pricing）：
+    // M3「永久五折」后 ¥2.1/¥8.4，与 M2.7 同价、比 M2.x highspeed 便宜一半；M2.5 / M2.1 / M2 官方已归入
+    // 「历史模型」。端点仍可用，`/model` 照收，pricing.ts 保留费率供历史回算。
   ]
 };
 var MiniMaxProvider = createStaticProviderClass(MINIMAX_CONFIG);
@@ -2069,14 +2071,20 @@ var MiniMaxProvider = createStaticProviderClass(MINIMAX_CONFIG);
 var QWEN_CONFIG = {
   id: "qwen",
   displayName: "Qwen (Alibaba Cloud)",
-  defaultModel: "qwen-plus",
+  defaultModel: "qwen3.8-flash",
   apiKeyEnvVar: "AICLI_API_KEY_QWEN",
   baseUrl: "https://dashscope.aliyuncs.com/compatible-mode/v1",
+  // 2026-10-18 按百炼官方文档更新（help.aliyun.com/zh/model-studio/models、/model-pricing、/deep-thinking）。
+  // **未经真实端点实测**（本机无 Qwen key）。三款 qwen3.x 默认开启思考、支持非流式调用、纯文本输入；
+  // 思考与非思考输出同价。上下文按官方价格阶梯的上限 1M 记。
   models: [
-    { id: "qwen-plus", displayName: "Qwen Plus", contextWindow: 131072, supportsStreaming: true },
-    { id: "qwen-max", displayName: "Qwen Max", contextWindow: 131072, supportsStreaming: true },
-    { id: "qwen-turbo", displayName: "Qwen Turbo", contextWindow: 131072, supportsStreaming: true },
-    { id: "qwen-coder-plus", displayName: "Qwen Coder Plus", contextWindow: 131072, supportsStreaming: true }
+    { id: "qwen3.8-max", displayName: "Qwen3.8 Max\uFF08\u65D7\u8230\uFF0C1M\uFF0C\xA512/\xA536\uFF09", contextWindow: 1e6, supportsStreaming: true },
+    { id: "qwen3.7-plus", displayName: "Qwen3.7 Plus\uFF081M\uFF0C\u2264256K \xA52/\xA58\uFF09", contextWindow: 1e6, supportsStreaming: true },
+    { id: "qwen3.8-flash", displayName: "Qwen3.8 Flash\uFF081M\uFF0C\xA50.8/\xA52.7\uFF09", contextWindow: 1e6, supportsStreaming: true },
+    // 老模型里唯一有价格优势的：≤128K ¥0.15/¥1.5，最便宜。默认不开思考。
+    { id: "qwen-flash", displayName: "Qwen Flash\uFF08\u6700\u4FBF\u5B9C\uFF0C\u2264128K \xA50.15/\xA51.5\uFF09", contextWindow: 1e6, supportsStreaming: true }
+    // 下架（端点若仍可用，`/model` 照收）：qwen-max ¥2.4/¥9.6 比 qwen3.7-plus 还贵；qwen-plus 价格不如
+    // qwen-flash、能力不如 qwen3.8-flash；qwen-turbo / qwen-coder-plus 已不在官方价格页上。
   ]
 };
 var QwenProvider = createStaticProviderClass(QWEN_CONFIG);

package/dist/{chunk-VKQIHFI5.js → chunk-GWEBO4RF.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   TEST_TIMEOUT
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 import {
   getExecutionConfigDir,
   getExecutionSignal

package/dist/{chunk-B4EBCSKM.js → chunk-HMQA2R3O.js}

@@ -8,7 +8,7 @@ import {
 import {
   CONFIG_DIR_NAME,
   PLUGINS_DIR_NAME
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/diagnostics/tool-stats.ts
 import { mkdirSync } from "fs";

package/dist/{chunk-WDEO7J3T.js → chunk-LLXHA2LA.js}

@@ -2,7 +2,7 @@
 import {
   computeCost,
   formatCost
-} from "./chunk-BRZZ3QAI.js";
+} from "./chunk-2BAFLDDL.js";
 import {
   clearRecentlyDeniedAutoActions,
   estimateTokens,
@@ -16,7 +16,7 @@ import {
   setPreferredAgentName,
   truncateForPersist,
   undoStack
-} from "./chunk-UQI5TOL2.js";
+} from "./chunk-M5B6G6AY.js";
 import {
   buildReviewPrompt,
   buildSecurityReviewPrompt
@@ -51,7 +51,7 @@ import {
   touchMemoryReferences,
   updateMemoryApproval,
   updateMemoryEntry
-} from "./chunk-E3RJEQZA.js";
+} from "./chunk-DX2U6QK4.js";
 import {
   theme
 } from "./chunk-5C76XDDF.js";
@@ -65,7 +65,7 @@ import {
 import {
   buildDoctorReport,
   formatDoctorReport
-} from "./chunk-IEKVRC2E.js";
+} from "./chunk-265M6ZLH.js";
 import {
   describePlugin,
   getActivePluginAssets,
@@ -80,7 +80,7 @@ import {
   trustHook,
   trustPlugin,
   untrustHook
-} from "./chunk-B4EBCSKM.js";
+} from "./chunk-HMQA2R3O.js";
 import {
   isCleanDocumentBody,
   stripOuterCodeFence
@@ -109,7 +109,7 @@ import {
   MCP_TOOL_PREFIX,
   REPO_URL,
   VERSION
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/skills/manager.ts
 import { existsSync, readdirSync, mkdirSync, statSync } from "fs";
@@ -2445,7 +2445,7 @@ No tools match "${filter}".
           const { join: join6 } = await import("path");
           const { existsSync: existsSync6 } = await import("fs");
           const { getGitRoot } = await import("./git-context-EXOEHQSF.js");
-          const { MCP_PROJECT_CONFIG_NAME } = await import("./constants-5YWODQHN.js");
+          const { MCP_PROJECT_CONFIG_NAME } = await import("./constants-DLJECTGQ.js");
           const { approveProject, hashMcpFile } = await import("./project-trust-KGG2BWKC.js");
           const cwd = process.cwd();
           const projectRoot = getGitRoot(cwd) ?? cwd;
@@ -3553,7 +3553,7 @@ function createReviewCommands() {
       usage: "/test [command|filter]",
       async execute(args, ctx) {
         try {
-          const { executeTests } = await import("./run-tests-CC24OLYI.js");
+          const { executeTests } = await import("./run-tests-H3DIUCDB.js");
           const argStr = args.join(" ").trim();
           let testArgs = {};
           if (argStr) {

package/dist/{chunk-U6VATMAI.js → chunk-LUJ4JVMI.js}

@@ -2,7 +2,7 @@
 import {
   evaluateCiGate,
   normalizeCiThresholds
-} from "./chunk-T4K4RZKR.js";
+} from "./chunk-CCFCPAKN.js";
 import {
   buildReviewPrompt,
   buildSecurityReviewPrompt,
@@ -13,13 +13,13 @@ import {
 } from "./chunk-6YZEZFX5.js";
 import {
   ProviderRegistry
-} from "./chunk-NKTCITLH.js";
+} from "./chunk-GEC3MEVQ.js";
 import {
   ConfigManager
-} from "./chunk-OVEZ6HB5.js";
+} from "./chunk-QU74ONFF.js";
 import {
   VERSION
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/cli/ci.ts
 import { execFileSync } from "child_process";

package/dist/{chunk-UQI5TOL2.js → chunk-M5B6G6AY.js}

@@ -11,7 +11,7 @@ import {
   searchPersistentMemories,
   touchMemoryReferences,
   updateMemoryEntry
-} from "./chunk-E3RJEQZA.js";
+} from "./chunk-DX2U6QK4.js";
 import {
   indexProject
 } from "./chunk-RWRZVQCB.js";
@@ -30,7 +30,7 @@ import {
   runHook,
   runLifecycleHooks,
   runTool
-} from "./chunk-B4EBCSKM.js";
+} from "./chunk-HMQA2R3O.js";
 import {
   buildTruncatedArgumentsError,
   getDangerLevel,
@@ -64,7 +64,7 @@ import {
   recoverGbkMojibake,
   runTestsTool,
   shellInvocation
-} from "./chunk-VKQIHFI5.js";
+} from "./chunk-GWEBO4RF.js";
 import {
   CONFIG_DIR_NAME,
   DEFAULT_MAX_TOOL_OUTPUT_CHARS_CAP,
@@ -73,7 +73,7 @@ import {
   SUBAGENT_ALLOWED_TOOLS,
   SUBAGENT_DEFAULT_MAX_ROUNDS,
   SUBAGENT_MAX_ROUNDS_LIMIT
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 import {
   getCurrentSessionKey,
   getExecutionConfigDir,

package/dist/{chunk-OVEZ6HB5.js → chunk-QU74ONFF.js}

@@ -11,7 +11,7 @@ import {
   CONFIG_FILE_NAME,
   HISTORY_DIR_NAME,
   PLUGINS_DIR_NAME
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 
 // src/config/config-manager.ts
 import { readFileSync, existsSync, mkdirSync } from "fs";
@@ -388,8 +388,14 @@ ${err}`
       "deepseek-v4-flash-vision-exp"
     ]);
     const DEEPSEEK_TARGET = "deepseek-flash";
-    const LEGACY_KIMI = /* @__PURE__ */ new Set(["kimi-k2-0711-preview", "kimi-k2-turbo-preview"]);
-    const KIMI_TARGET = "kimi-k2.6";
+    const LEGACY_KIMI = /* @__PURE__ */ new Set([
+      "kimi-k2-0711-preview",
+      "kimi-k2-turbo-preview",
+      "kimi-k2.5",
+      "kimi-k2-thinking",
+      "kimi-k2-thinking-turbo"
+    ]);
+    const KIMI_TARGET = "kimi-k2.7-code";
     const LEGACY_CLAUDE = {
       "claude-sonnet-4-5-20250929": "claude-sonnet-5-5",
       "claude-sonnet-4-5": "claude-sonnet-5-5",

package/dist/{ci-U5Q3BMQP.js → ci-6K3JNHB4.js}

@@ -3,16 +3,16 @@ import {
   CI_COMMENT_MARKER,
   countSeverity,
   runCi
-} from "./chunk-U6VATMAI.js";
-import "./chunk-T4K4RZKR.js";
+} from "./chunk-LUJ4JVMI.js";
+import "./chunk-CCFCPAKN.js";
 import "./chunk-HLWUDRBO.js";
 import "./chunk-6YZEZFX5.js";
-import "./chunk-NKTCITLH.js";
-import "./chunk-OVEZ6HB5.js";
+import "./chunk-GEC3MEVQ.js";
+import "./chunk-QU74ONFF.js";
 import "./chunk-E2W2MI6H.js";
 import "./chunk-TU3L3PW7.js";
 import "./chunk-BMLQPSHP.js";
-import "./chunk-NOQKCHIQ.js";
+import "./chunk-AJITHMRE.js";
 export {
   CI_COMMENT_MARKER,
   countSeverity,

package/dist/{ci-format-4RR2G3RR.js → ci-format-2L3RFJVH.js}

@@ -6,8 +6,8 @@ import {
   formatCiResult,
   formatCiSarif,
   normalizeCiThresholds
-} from "./chunk-T4K4RZKR.js";
-import "./chunk-NOQKCHIQ.js";
+} from "./chunk-CCFCPAKN.js";
+import "./chunk-AJITHMRE.js";
 export {
   DEFAULT_CI_THRESHOLDS,
   detectCiGateSignals,

package/dist/{constants-5YWODQHN.js → constants-DLJECTGQ.js}

@@ -39,7 +39,7 @@ import {
   TEST_TIMEOUT,
   VERSION,
   buildUserIdentityPrompt
-} from "./chunk-NOQKCHIQ.js";
+} from "./chunk-AJITHMRE.js";
 export {
   AGENTIC_BEHAVIOR_GUIDELINE,
   APP_NAME,

package/dist/{constants-J3346CYX.js → constants-J5SJ232X.js}

@@ -38,7 +38,7 @@ import {
   TEST_TIMEOUT,
   VERSION,
   buildUserIdentityPrompt
-} from "./chunk-LOS7T2JS.js";
+} from "./chunk-3YVZJHTF.js";
 export {
   AGENTIC_BEHAVIOR_GUIDELINE,
   APP_NAME,