responses 直通超时可配,流式去总超时,代理补阶段超时

docs/docs.go

@@ -860,7 +860,7 @@ const docTemplate = `{
                 "summary": "更新 AI 网关全局设置",
                 "parameters": [
                     {
-                        "description": "全量提交;保险丝阈值限 1..1024 KB",
+                        "description": "全量提交;保险丝阈值限 1..1024 KB,上游无响应预算限 30..900 秒",
                         "name": "body",
                         "in": "body",
                         "required": true,
@@ -6102,6 +6102,10 @@ const docTemplate = `{
                 },
                 "streamGuardKB": {
                     "type": "integer"
+                },
+                "upstreamWaitSeconds": {
+                    "description": "UpstreamWaitSeconds 是 responses 直通的上游无响应预算(秒):非流式为单次\n尝试总超时,流式为等待响应头预算;multi-agent/搜索类模型需远超 60s",
+                    "type": "integer"
                 }
             }
         },

docs/swagger.json

@@ -853,7 +853,7 @@
                 "summary": "更新 AI 网关全局设置",
                 "parameters": [
                     {
-                        "description": "全量提交;保险丝阈值限 1..1024 KB",
+                        "description": "全量提交;保险丝阈值限 1..1024 KB,上游无响应预算限 30..900 秒",
                         "name": "body",
                         "in": "body",
                         "required": true,
@@ -6095,6 +6095,10 @@
                 },
                 "streamGuardKB": {
                     "type": "integer"
+                },
+                "upstreamWaitSeconds": {
+                    "description": "UpstreamWaitSeconds 是 responses 直通的上游无响应预算(秒):非流式为单次\n尝试总超时,流式为等待响应头预算;multi-agent/搜索类模型需远超 60s",
+                    "type": "integer"
                 }
             }
         },

docs/swagger.yaml

@@ -65,6 +65,11 @@ definitions:
         type: boolean
       streamGuardKB:
         type: integer
+      upstreamWaitSeconds:
+        description: |-
+          UpstreamWaitSeconds 是 responses 直通的上游无响应预算(秒):非流式为单次
+          尝试总超时,流式为等待响应头预算;multi-agent/搜索类模型需远超 60s
+        type: integer
     type: object
   internal_api.attachBootVolumeRequest:
     properties:
@@ -3204,7 +3209,7 @@ paths:
       - AI 管理
     put:
       parameters:
-      - description: 全量提交;保险丝阈值限 1..1024 KB
+      - description: 全量提交;保险丝阈值限 1..1024 KB,上游无响应预算限 30..900 秒
         in: body
         name: body
         required: true