diff --git a/llmzoo/eval/eval_gpt_review_all.py b/llmzoo/eval/eval_gpt_review_all.py
index 34cff04..6a297fa 100644
--- a/llmzoo/eval/eval_gpt_review_all.py
+++ b/llmzoo/eval/eval_gpt_review_all.py
@@ -6,13 +6,14 @@
 import backoff
 import numpy as np
 import openai
+import litellm
 import ray
 
 
 @ray.remote(num_cpus=4)
 @backoff.on_exception(backoff.expo, openai.error.RateLimitError)
 def get_eval(content: str, max_tokens: int):
-    response = openai.ChatCompletion.create(
+    response = litellm.completion(
         model='gpt-3.5-turbo',
         messages=[{
             'role': 'system',
diff --git a/llmzoo/eval/prompt_turbo.py b/llmzoo/eval/prompt_turbo.py
index b426d81..f5bc6bf 100644
--- a/llmzoo/eval/prompt_turbo.py
+++ b/llmzoo/eval/prompt_turbo.py
@@ -3,6 +3,7 @@
 
 import backoff
 import openai
+import litellm
 import ray
 import shortuuid
 
@@ -12,7 +13,7 @@
 @ray.remote(num_cpus=4)
 @backoff.on_exception(backoff.expo, openai.error.RateLimitError)
 def completions_with_backoff(**kwargs):
-    return openai.ChatCompletion.create(**kwargs)
+    return litellm.completion(**kwargs)
 
 
 if __name__ == '__main__':