model is now enviromental variable LLM_MODEL

app.py

@@ -6,6 +6,14 @@ import gradio as gr
 import time
 import re
 #export GRADIO_DEBUG=1
+# ----------- CONFIGURATION ----------------------------------------------------
+# OPENAI_API_KEY must be set in the environment
+
+# Model name for LLM calls. Can be overridden by setting the LLM_MODEL
+# environment variable. Falls back to a sensible default if unset.
+MODEL_NAME = os.getenv("LLM_MODEL", "models/gemini-flash-latest")
+
+# LLM_MODEL_NAME must be set in the environment
 
 def search_inspire(query, size=10):
     """
@@ -71,7 +79,7 @@ def llm_expand_query(query):
   """ Expands a query to variations of fulltext searches """
 
   response = client.chat.completions.create(
-    model="gpt-4o-mini",
+    model=MODEL_NAME,
     messages=[
       {
         "role": "user",
@@ -120,7 +128,7 @@ def llm_generate_answer(prompt):
   """ Generate a response from the LLM """
 
   response = client.chat.completions.create(
-    model="gpt-4o-mini",
+    model=MODEL_NAME,
     messages=[
       {
         "role": "system",