Fix: Broken DashScope support introduced in PR 441 (previous integration was non‑native and unusable)

frontend/src/app/setting/components/models/model-detail.tsx

@@ -135,6 +135,8 @@ export function ModelDetail({ provider }: ModelDetailProps) {
     settingDefaultModel ||
     settingDefaultProvider;
 
+  const isDashScope = (provider ?? "").toLowerCase() === "dashscope";
+
   if (detailLoading) {
     return (
       <div className="text-gray-400 text-sm">Loading provider details...</div>
@@ -216,23 +218,24 @@ export function ModelDetail({ provider }: ModelDetailProps) {
               )}
             </configForm.Field>
 
-            {/* API Host section */}
-            <configForm.Field name="base_url">
-              {(field) => (
-                <Field className="text-gray-950">
-                  <FieldLabel className="font-medium text-base">
-                    API Host
-                  </FieldLabel>
-                  <Input
-                    placeholder={providerDetail.base_url}
-                    value={field.state.value}
-                    onChange={(e) => field.handleChange(e.target.value)}
-                    onBlur={() => configForm.handleSubmit()}
-                  />
-                  <FieldError errors={field.state.meta.errors} />
-                </Field>
-              )}
-            </configForm.Field>
+            {!isDashScope && (
+              <configForm.Field name="base_url">
+                {(field) => (
+                  <Field className="text-gray-950">
+                    <FieldLabel className="font-medium text-base">
+                      API Host
+                    </FieldLabel>
+                    <Input
+                      placeholder={providerDetail.base_url}
+                      value={field.state.value}
+                      onChange={(e) => field.handleChange(e.target.value)}
+                      onBlur={() => configForm.handleSubmit()}
+                    />
+                    <FieldError errors={field.state.meta.errors} />
+                  </Field>
+                )}
+              </configForm.Field>
+            )}
           </FieldGroup>
 
           {/* Models section */}

python/configs/providers/dashscope.yaml

@@ -1,85 +1,90 @@
 # ============================================
 # DashScope Provider Configuration
 # ============================================
-# DashScope (Alibaba Cloud Bailian) exposes an OpenAI-compatible API for Qwen3 models.
-# Configure the API key via DASHSCOPE_API_KEY or override using env vars at runtime.
+name: "DashScope"
+provider_type: "dashscope"
 
-name: DashScope
-provider_type: dashscope
-enabled: true
+enabled: true  # Default is true if not specified
 
-# Connection parameters for DashScope compatible-mode endpoint.
+# Connection Configuration
 connection:
-  base_url: https://dashscope.aliyuncs.com/compatible-mode/v1
-  api_key_env: DASHSCOPE_API_KEY
+  # DashScope (Alibaba Cloud Bailian) native HTTP uses fixed official host.
+  # base_url is not required; leave empty unless using a proxy.
+  api_key_env: "DASHSCOPE_API_KEY"
+  base_url: ""
 
-# Default chat model used when no model_id is specified.
-default_model: qwen3-max
+# Default chat model if none specified
+default_model: "qwen3-max"
 
-# Global default inference parameters.
+# Model Parameters Defaults
 defaults:
   temperature: 0.7
   max_tokens: 16384
 
-# Commonly used Qwen3 models available via DashScope.
+# Available Models
 models:
-  - id: qwen3-max
-    name: Qwen3 Max
+  - id: "qwen3-max"
+    name: "Qwen3 Max"
     context_length: 256000
     max_output_tokens: 16384
-    description: Qwen3 Max model with strongest performance
+    description: "Qwen3 Max model with strongest performance"
     supported_inputs:
       - text
     supported_outputs:
       - text
 
-  - id: qwen3-max-preview
-    name: Qwen3 Max Preview
+  - id: "qwen3-max-preview"
+    name: "Qwen3 Max Preview"
     context_length: 256000
     max_output_tokens: 16384
-    description: Qwen3 Max preview model
+    description: "Qwen3 Max preview model"
     supported_inputs:
       - text
     supported_outputs:
       - text
 
-  - id: qwen-plus
-    name: Qwen Plus
+  - id: "qwen-plus"
+    name: "Qwen Plus"
     context_length: 256000
     max_output_tokens: 16384
-    description: Qwen Plus model with balanced performance
+    description: "Qwen Plus model with balanced performance"
     supported_inputs:
       - text
     supported_outputs:
       - text
 
-  - id: qwen-flash
-    name: Qwen Flash
+  - id: "qwen-flash"
+    name: "Qwen Flash"
     context_length: 256000
     max_output_tokens: 16384
-    description: Qwen Flash model optimized for fast response
+    description: "Qwen Flash model optimized for fast response"
     supported_inputs:
       - text
     supported_outputs:
       - text
 
-# Embedding configuration for DashScope text embedding models.
+# ============================================
+# Embedding Models Configuration
+# ============================================
 embedding:
-  default_model: text-embedding-v4
+  # Default embedding model
+  default_model: "text-embedding-v4"
 
+  # Default parameters
   defaults:
     dimensions: 2048
     encoding_format: "float"
 
+  # Available embedding models
   models:
-    - id: text-embedding-v4
-      name: Text Embedding V4
+    - id: "text-embedding-v4"
+      name: "Text Embedding V4"
       dimensions: 2048
       max_input: 8192
-      description: DashScope text embedding v4 model (latest)
+      description: "DashScope text embedding v4 model (latest)"
 
-    - id: text-embedding-v3
-      name: Text Embedding V3
+    - id: "text-embedding-v3"
+      name: "Text Embedding V3"
       dimensions: 1024
       max_input: 8192
-      description: DashScope text embedding v3 model
+      description: "DashScope text embedding v3 model"

python/valuecell/adapters/models/__init__.py

@@ -30,6 +30,8 @@
     OpenAIProvider,
     OpenRouterProvider,
     SiliconFlowProvider,
+    create_embedder,
+    create_embedder_for_agent,
     create_model,
     create_model_for_agent,
     get_model_factory,
@@ -52,4 +54,6 @@
     # Convenience functions
     "create_model",
     "create_model_for_agent",
+    "create_embedder",
+    "create_embedder_for_agent",
 ]