diff --git a/litellm/__init__.py b/litellm/__init__.py
index e3d4a8d849f..6ae281afc9f 100644
--- a/litellm/__init__.py
+++ b/litellm/__init__.py
@@ -2,7 +2,7 @@ success_callback = []
 failure_callback = []
 set_verbose=False
 telemetry=True
-
+max_tokens = 256 # OpenAI Defaults
 ####### PROXY PARAMS ################### configurable params if you use proxy models like Helicone
 api_base = None
 headers = None
diff --git a/litellm/main.py b/litellm/main.py
index 16faba2bd6f..898f40ae416 100644
--- a/litellm/main.py
+++ b/litellm/main.py
@@ -189,7 +189,7 @@ def completion(
       if max_tokens != float('inf'):
         max_tokens_to_sample = max_tokens
       else:
-        max_tokens_to_sample = 300 # default in Anthropic docs https://docs.anthropic.com/claude/reference/client-libraries
+        max_tokens_to_sample = litellm.max_tokens # default in Anthropic docs https://docs.anthropic.com/claude/reference/client-libraries
       ## LOGGING
       logging(model=model, input=prompt, azure=azure, additional_args={"max_tokens": max_tokens}, logger_fn=logger_fn)
       ## COMPLETION CALL