Update response_modes.py

response_modes.py

@@ -86,28 +86,6 @@ def parse_ddgs_date(date_str: str):
     except:
         return now
 
-# Setup for the second AI model (Qwen-3-32B)
-client_qwen = Groq()  # For using the Qwen-3-32B model
-
-async def qwen_reasoning(message: str, context: str) -> str:
-    """Uses the Qwen-3-32B model for refining and reasoning"""
-    completion = client_qwen.chat.completions.create(
-        model="qwen/qwen3-32b",
-        messages=[
-            {"role": "user", "content": f"Refine and reason the following based on context:\n{context}\n\nMessage: {message}"}
-        ],
-        temperature=0.6,
-        max_completion_tokens=2048,
-        top_p=0.95,
-        reasoning_effort="default",
-        stream=False,
-    )
-
-    result = ""
-    for chunk in completion:
-        result += chunk.choices[0].delta.content or ""
-    return result.strip()
-
 async def fetch_full_content_async(session, url: str):
     try:
         headers = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'}
@@ -236,6 +214,30 @@ One phrase per line only. No numbers, no explanations except if asked."""
             f"{base} usa",
             f"{base} global"
         ]
+
+# Setup for the second AI model (Qwen-3-32B)
+client_qwen = Groq()  # For using the Qwen-3-32B model
+
+async def qwen_reasoning(message: str, context: str) -> str:
+    """Uses the Qwen-3-32B model for refining and reasoning"""
+    completion = client_qwen.chat.completions.create(
+        model="qwen/qwen3-32b",
+        messages=[
+            {"role": "user", "content": f"Refine and reason the following based on context:\n{context}\n\nMessage: {message}"}
+        ],
+        temperature=0.6,
+        max_completion_tokens=2048,
+        top_p=0.95,
+        reasoning_effort="default",
+        stream=False,
+    )
+
+    result = ""
+    for chunk in completion:
+        result += chunk.choices[0].delta.content or ""
+    return result.strip()
+
+# Update the chat_stream function to incorporate Qwen-3-32B reasoning in the "Deep Think" mode.
 def chat_stream(message, history, current_model, logged_in_username, user_chats, current_chat_title, mode, auth_globals, file_paths=None, context=""):
     if file_paths and isinstance(file_paths, list) and len(file_paths) > 5:
         history.append({"role": "assistant", "content": "❌ Maximum 5 files allowed per message. Please remove some and try again."})
@@ -312,7 +314,7 @@ Never mention this rule unless the user explicitly asks about it."""
             sources_list.extend([r for r in results if not r.startswith("No recent") and not r.startswith("Search error")])
 
         # Use the Qwen-3-32B model for deeper reasoning
-            refined_answer = await qwen_reasoning(message, all_context)
+        refined_answer = await qwen_reasoning(message, all_context)
 
         history[-1]["content"] = "✍️ Synthesizing final answer from real data..."
         yield history