Adding Google Gemini 2.5 flash lite

Config_files/message_system_config.json

@@ -21,12 +21,13 @@
   "AI_phrases_singeo": ["your voice deserves more"],
   "header_limit": 30,
   "message_limit": 110,
-  "LLM_models": ["gpt-4o-mini", "gpt-4o", "gpt-4.1-mini", "gpt-3.5-turbo", "o4-mini", "o1-mini", "o3-mini", "claude-3-5-haiku-latest", "google/gemma-3-27b-instruct/bf-16", "meta-llama/llama-3.2-11b-instruct/fp-16"],
+  "LLM_models": ["gpt-4o-mini", "gpt-4o", "gpt-4.1-mini", "gpt-3.5-turbo", "o4-mini", "o1-mini", "o3-mini", "claude-3-5-haiku-latest", "google/gemma-3-27b-instruct/bf-16", "meta-llama/llama-3.2-11b-instruct/fp-16", "gemini-2.5-flash-lite-preview-06-17"],
   "openai_models": ["gpt-4o-mini", "gpt-4o", "gpt-4.1-nano", "gpt-3.5-turbo", "gpt-4.1-mini"],
   "reasoning": ["o1", "o4-mini", "o1-mini", "o3-mini"],
   "ollama_models": ["deepseek-r1:1.5b", "gemma3:4b", "deepseek-r1:7b", "gemma3:4b"],
   "claude_models": ["claude-3-5-haiku-latest"],
-  "inference_models": ["google/gemma-3-27b-instruct/bf-16", "meta-llama/llama-3.2-11b-instruct/fp-16"]
+  "inference_models": ["google/gemma-3-27b-instruct/bf-16", "meta-llama/llama-3.2-11b-instruct/fp-16"],
+  "google_models": ["gemini-2.5-flash-lite-preview-06-17"]
 }
 
 

Messaging_system/LLM.py

@@ -12,6 +12,8 @@ import re
 import anthropic
 import os
 import streamlit as st
+from google.genai import types
+from google import genai
 
 
 
@@ -35,6 +37,8 @@ class LLM:
             response = self.get_message_inference(prompt, instructions)
         elif self.model_type == "claude":
             response = self.get_message_claude(prompt, instructions)
+        elif self.model_type == "google":
+            response = self.get_message_google(prompt, instructions)
         else:
             raise f"Invalid model type : {self.model_type}"
 
@@ -49,9 +53,11 @@ class LLM:
         if self.Core.model in self.Core.config_file["openai_models"]:
             self.model_type = "openai"
 
-        if self.Core.model in self.Core.config_file["inference_models"]:
+        elif self.Core.model in self.Core.config_file["inference_models"]:
             self.model_type = "inference"
 
+        elif self.Core.model in self.Core.config_file["google_models"]:
+            self.model_type = "google"
 
         elif self.Core.model in self.Core.config_file["ollama_models"]:
             self.model_type = "ollama"
@@ -149,6 +155,42 @@ class LLM:
         print("Max retries exceeded. Returning empty response.")
         return None
 
+    # =========================================================================
+    def get_message_google(self, prompt, instructions, max_retries=4):
+
+        client = genai.Client(api_key=self.get_credential("Google_API"))
+
+        for attempt in range(max_retries):
+            try:
+                response = client.models.generate_content(
+                    model=self.Core.model,
+                    contents=prompt,
+                    config=types.GenerateContentConfig(
+                        thinking_config=types.ThinkingConfig(thinking_budget=0),
+                        system_instruction=instructions,
+                        temperature=self.Core.temperature,
+                        response_mime_type="application/json"
+                    ))
+
+                output = json.loads(str(response.text))
+
+                if 'message' not in output or 'header' not in output:
+                    print(f"'message' or 'header' is missing in response on attempt {attempt + 1}. Retrying...")
+                    continue  # Continue to next attempt
+
+                else:
+                    if len(output["header"].strip()) > self.Core.config_file["header_limit"] or len(
+                            output["message"].strip()) > self.Core.config_file["message_limit"]:
+                        print(
+                            f"'header' or 'message' is more than specified characters in response on attempt {attempt + 1}. Retrying...")
+                        continue
+                return output
+
+            except json.JSONDecodeError:
+                print(f"Invalid JSON from LLM on attempt {attempt + 1}. Retrying...")
+            except Exception as e:
+                print(f"Error in attempt {attempt}: {e}")
+
     # =========================================================================
 
     def get_message_openai(self, prompt, instructions, max_retries=4):

Messaging_system/PromptEng.py

@@ -7,6 +7,8 @@ from openai import OpenAI
 from Messaging_system.LLM import LLM
 import os
 import streamlit as st
+from google.genai import types
+from google import genai
 
 class PromptEngine:
 
@@ -52,7 +54,7 @@ output the new prompt as text without any additional information.
             final_prompt = self.get_openai_response(prompt)
             return final_prompt
 
-        if self.Core.model in self.Core.config_file["inference_models"]:
+        elif self.Core.model in self.Core.config_file["inference_models"]:
             final_prompt = self.get_inference_response(prompt)
             return final_prompt
 
@@ -60,6 +62,10 @@ output the new prompt as text without any additional information.
             final_prompt = self.get_claude_response(prompt, self.llm_instructions())
             return final_prompt
 
+        elif self.Core.model in self.Core.config_file["google_models"]:
+            final_prompt = self.get_gemini_response(prompt)
+            return final_prompt
+
     # ============================================================
     def llm_instructions(self):
 
@@ -199,6 +205,34 @@ output the new prompt as text without any additional information.
         return prompt # returns original prompt if needed
 
     # ==========================================================================
+    def get_gemini_response(self, prompt, max_retries=4):
+        """
+        Send prompt to Google Gemini LLM and get back the response
+        :param prompt:
+        :param max_retries:
+        :return:
+        """
+
+        client = genai.Client(api_key=self.get_credential("Google_API"))
+
+        for attempt in range(max_retries):
+            try:
+                response = client.models.generate_content(
+                    model=self.Core.model,
+                    contents=prompt,
+                    config=types.GenerateContentConfig(
+                        thinking_config=types.ThinkingConfig(thinking_budget=0),
+                        system_instruction=self.llm_instructions(),
+                        temperature=self.Core.temperature,
+                        response_mime_type = "text/plain"  # application/json
+                    ))
+
+                output = str(response.text)
+                return output
+            except Exception as e:
+                print(f"Error in attempt {attempt}: {e}")
+
+    # ==========================================================================
 
 
     def get_claude_response(self, prompt, instructions, max_retries=4):
@@ -231,3 +265,4 @@ output the new prompt as text without any additional information.
         print("Max retries exceeded. Returning empty response.")
         return prompt # returns original prompt if needed
 
+

messaging_main_test.py

@@ -163,7 +163,7 @@ if __name__ == "__main__":
     # o3-mini o1-mini o4-mini o1
 
     users_message = permes.create_personalize_messages(session=session,
-                                                model="osmosis-ai/osmosis-structure-0.6b/fp-32",
+                                                model="gemini-2.5-flash-lite-preview-06-17",
                                                 users=users,
                                                 brand=brand,
                                                 config_file=config_file,