@@ -429,7 +429,7 @@ Key variables below; the full list is in `src/config.py` (authoritative), and **
 | `USERS_FILE` | `secrets/users.txt` | **Bootstrap-only**: legacy user file imported once (idempotent) |
 | `PUBLIC_BASE_URL` | `http://127.0.0.1:8000` | Browser-reachable address; TRAE login callback depends on it |
 | `DATA_DIR` | `./data` | SQLite and runtime data directory |
-| `DEFAULT_MODEL` | `glm-5.2` | Model when empty/`auto` |
+| `DEFAULT_MODEL` | `deepseek-v4.1-flash` | Model when empty/`auto` |
 | `QUOTA_PROBE_MINUTES` | `60` | Quota probe interval (floor 1) |
 | `REFRESH_INTERVAL_MINUTES` | `30` | Token pre-refresh polling interval (floor 5); short-lived channels cap their window wider than this |
 | `MODEL_CATALOG_MINUTES` | `30` | Channel catalog fallback refresh (floor 5) |
 
@@ -602,7 +602,7 @@ python3 scripts/restore_db.py data/backups/coding2api-20261014-030000.sqlite3 --
 | `USERS_FILE` | `secrets/users.txt` | **仅引导期**：老式用户文件路径，启动时一次性导入 SQLite（幂等）。账号唯一源是 `users` 表 |
 | `PUBLIC_BASE_URL` | `http://127.0.0.1:8000` | 浏览器可达地址；TRAE 登录回调（`+/authorize`）依赖它，scheme 为 https 时额外下发 HSTS。**不决定会话 Cookie 的 `Secure`**——那按请求实际协议动态判定（`X-Forwarded-Proto` 优先，无该头则看 Host），同一实例 LAN(http) 与隧道(https) 双入口才能都登录 |
 | `DATA_DIR` | `./data` | SQLite 与运行数据目录 |
-| `DEFAULT_MODEL` | `glm-5.2` | 模型为空/`auto` 时的默认 |
+| `DEFAULT_MODEL` | `deepseek-v4.1-flash` | 模型为空/`auto` 时的默认 |
 | `QUOTA_PROBE_MINUTES` | `60` | 额度探测周期（下限 1 分钟） |
 | `REFRESH_INTERVAL_MINUTES` | `30` | token 预刷新轮询周期（下限 5 分钟）；短寿命渠道的预刷新窗口封顶必须**宽于**它 |
 | `MODEL_CATALOG_MINUTES` | `30` | 渠道模型列表兜底刷新周期（下限 5 分钟） |
 
@@ -18,7 +18,7 @@ services:
       # 远程部署时必须设为浏览器可达的地址，TRAE 登录回调依赖它
       PUBLIC_BASE_URL: ${PUBLIC_BASE_URL:-http://127.0.0.1:8000}
       CODEBUDDY_API_ENDPOINT: ${CODEBUDDY_API_ENDPOINT:-https://copilot.tencent.com}
-      DEFAULT_MODEL: ${DEFAULT_MODEL:-glm-5.2}
+      DEFAULT_MODEL: ${DEFAULT_MODEL:-deepseek-v4.1-flash}
       # 模型列表黑名单（fnmatch glob，逗号分隔，完全替换语义）
       MODEL_BLOCKLIST: ${MODEL_BLOCKLIST:-custom_model_*,*sub*agent*,summary,browser_use_*,file_search_agent,default,seed-code-pro-*,sagitta,aquila,glm-4.*,glm-5,glm-5-turbo,glm-5.0*,glm-5.1,glm-5v-*,kimi-k2.5,kimi-k2.6,kimi-k2-thinking,minimax-m2.5,deepseek-v3-*,hunyuan-*,hy3*,deepseek-v4-*,qwen3.7-*,auto,nemotron-3-*,glm-5.2*,nemotron-3*,lfm2*,north-mini-code,openrouter-models-router,fast,balanced,deep,primary,hy4-preview,inkling-small,minimax-m2.7,kimi-k2.*,minimax-m3,gpt-5.4,gpt-5.5,laguna*,qwen3-vl-235b,seed-2.1*,xing4.0-29b-a4,glyph-cluster,qwen3.8-flash,seed-evolving}
       QUOTA_PROBE_MINUTES: ${QUOTA_PROBE_MINUTES:-60}
 
@@ -89,7 +89,9 @@ class Settings(BaseSettings):
     provider_proxies: str = ""
 
     # 路由与调度
-    default_model: str = "glm-5.2"
+    # 默认模型：六条渠道都已上架 `deepseek-v4.1-flash`（CB / TRAE / CodeArts /
+    # codebuddy_intl 同名，Qoder 走别名 `dfmodel`），故默认值取它而不是老模型
+    default_model: str = "deepseek-v4.1-flash"
     # token 预刷新的提前量（小时）：距 access token 到期 ≤ 该值就提前换新。
     # 注意它对**短寿命渠道**会被自行封顶（CodeArts STS 只有 2h，见
     # `CodeArtsCredential.refresh_skew_cap_seconds`），所以调大它不会让短寿命
 
@@ -12,7 +12,7 @@
 def test_settings_defaults_and_admin_set():
     s = Settings(_env_file=None, APP_SECRET=SECRET, ADMIN_USERNAMES="alice, bob ,")
     assert s.admin_set == frozenset({"alice", "bob"})
-    assert s.default_model == "glm-5.2"
+    assert s.default_model == "deepseek-v4.1-flash"
     assert s.quota_expiry_window_seconds == EXPIRY_WINDOW_SECONDS
     assert s.quota_expiry_secondary_window_seconds == SECONDARY_EXPIRY_WINDOW_SECONDS
     assert s.is_admin("alice") and not s.is_admin("carol")