Update app.py

app.py

@@ -49,7 +49,7 @@ class ChatOpenRouter(ChatOpenAI):
             **kwargs
         )
  
-llm = ChatOpenRouter(
+llm2 = ChatOpenRouter(
     #model_name="deepseek/deepseek-r1-0528:free",
     #model_name="google/gemini-2.0-flash-exp:free",
     #model_name="deepseek/deepseek-v3-base:free",
@@ -89,11 +89,11 @@ load_dotenv()
 # os.environ["GROQ_API_KEY"] = os.getenv("GROQ_API_KEY")
 groq_api_key = os.getenv("GROQ_API_KEY")
 
-# llm = ChatGroq(
-#     model="meta-llama/llama-4-scout-17b-16e-instruct",
-#     temperature=0,
-#     max_tokens=None,
-# )
+llm = ChatGroq(
+    model="meta-llama/llama-4-scout-17b-16e-instruct",
+    temperature=0,
+    max_tokens=None,
+)
 
 app = Flask(__name__)
 
@@ -267,6 +267,11 @@ agent = create_react_agent(
     prompt=SYSTEM_PROMPT
 )
 
+agent_2 = create_react_agent(
+    model=llm2,
+    tools=[], # No specific tools are defined here, but could be added later
+    prompt=SYSTEM_PROMPT
+)
 agent_json_resolver = create_react_agent(
     model=llm,
     tools=[], # No specific tools are defined here, but could be added later
@@ -1039,7 +1044,7 @@ Each plan must include a **single Scratch Hat Block** (e.g., 'event_whenflagclic
 """
 
     try:
-        response = agent.invoke({"messages": [{"role": "user", "content": planning_prompt}]})
+        response = agent_2.invoke({"messages": [{"role": "user", "content": planning_prompt}]})
         print("Raw response from LLM [OverallPlannerNode 1]:",response)
         raw_response = response["messages"][-1].content#strip_noise(response["messages"][-1].content)
         print("Raw response from LLM [OverallPlannerNode 2]:", raw_response) # Uncomment for debugging
@@ -1342,7 +1347,7 @@ Use sprite names exactly as provided in `sprite_names` (e.g., 'Sprite1', 'soccer
     - If feedback is minor, make precise, minimal improvements only.
 """
     try:
-        response = agent.invoke({"messages": [{"role": "user", "content": refinement_prompt}]})
+        response = agent_2.invoke({"messages": [{"role": "user", "content": refinement_prompt}]})
         raw_response = response["messages"][-1].content#strip_noise(response["messages"][-1].content)
         logger.info(f"Raw response from LLM [RefinedPlannerNode]: {raw_response[:500]}...")
         # json debugging and solving
@@ -1531,7 +1536,7 @@ Example output:
 ```
 """                                    
             try:
-                response = agent.invoke({"messages": [{"role": "user", "content": refinement_prompt}]})
+                response = agent_2.invoke({"messages": [{"role": "user", "content": refinement_prompt}]})
                 llm_output = response["messages"][-1].content
                 llm_json = extract_json_from_llm_response(llm_output)                 
                 logger.info(f"Successfully analyze the opcode requirement for {sprite} - {event}.")