feat: implement offline memory extraction

athena/app/services/llm_service.py

@@ -3,6 +3,7 @@
 from langchain_anthropic import ChatAnthropic
 from langchain_core.language_models.chat_models import BaseChatModel
 from langchain_google_genai import ChatGoogleGenerativeAI
+from langchain_google_vertexai import ChatVertexAI
 
 from athena.app.services.base_service import BaseService
 from athena.chat_models.custom_chat_openai import CustomChatOpenAI
@@ -50,6 +51,17 @@ def get_model(
             max_tokens_to_sample=max_output_tokens,
             max_retries=3,
         )
+    elif model_name.startswith("vertex:"):
+        # example: model_name="vertex:gemini-2.5-pro"
+        vertex_model = model_name.split("vertex:", 1)[1]
+        return ChatVertexAI(
+            model_name=vertex_model,
+            project="prometheus-code-agent",
+            location="us-central1",
+            temperature=temperature,
+            max_output_tokens=max_output_tokens,
+            max_retries=3,
+        )
     elif "gemini" in model_name:
         return ChatGoogleGenerativeAI(
             model=model_name,