caching with model names

litellm/__init__.py

@@ -20,6 +20,7 @@ vertex_location: Optional[str] = None
 hugging_api_token: Optional[str] = None
 togetherai_api_key: Optional[str] = None
 caching = False
+caching_with_models = False # if you want the caching key to be model + prompt
 model_cost = {
     "gpt-3.5-turbo": {
         "max_tokens": 4000,

litellm/tests/test_bad_params.py

@@ -50,4 +50,4 @@ try:
 except:
     print(f"error occurred: {traceback.format_exc()}")
     pass
-os.environ["OPENAI_API_KEY"] = str(temp_key) # this passes linting#5
+os.environ["OPENAI_API_KEY"] = str(temp_key)  # this passes linting#5

litellm/tests/test_caching.py

@@ -12,13 +12,13 @@ import pytest
 import litellm
 from litellm import embedding, completion
 
-litellm.caching = True
 messages = [{"role": "user", "content": "who is ishaan Github?  "}]
 
 
 # test if response cached
 def test_caching():
     try:
+        litellm.caching = True
         response1 = completion(model="gpt-3.5-turbo", messages=messages)
         response2 = completion(model="gpt-3.5-turbo", messages=messages)
         print(f"response1: {response1}")
@@ -32,3 +32,21 @@ def test_caching():
         litellm.caching = False
         print(f"error occurred: {traceback.format_exc()}")
         pytest.fail(f"Error occurred: {e}")
+
+
+
+def test_caching_with_models():
+    litellm.caching_with_models = True
+    response2 = completion(model="gpt-3.5-turbo", messages=messages)
+    response3 = completion(model="command-nightly", messages=messages)
+    print(f"response2: {response2}")
+    print(f"response3: {response3}")
+    litellm.caching_with_models = False
+    if response3 == response2:
+        # if models are different, it should not return cached response
+        print(f"response2: {response2}")
+        print(f"response3: {response3}")
+        pytest.fail(f"Error occurred: {e}")
+
+
+

litellm/utils.py

@@ -263,8 +263,17 @@ def client(original_function):
             if (
                 prompt != None and prompt in local_cache
             ):  # check if messages / prompt exists
-                result = local_cache[prompt]
-                return result
+                if litellm.caching_with_models:
+                    # if caching with model names is enabled, key is prompt + model name
+                    if (
+                        "model" in kwargs
+                        and kwargs["model"] in local_cache[prompt]["models"]
+                    ):
+                        cache_key = prompt + kwargs["model"]
+                        return local_cache[cache_key]
+                else:  # caching only with prompts
+                    result = local_cache[prompt]
+                    return result
             else:
                 return None
         except:
@@ -273,7 +282,15 @@ def client(original_function):
     def add_cache(result, *args, **kwargs):
         try:  # never block execution
             prompt = get_prompt(*args, **kwargs)
-            local_cache[prompt] = result
+            if litellm.caching_with_models:  # caching with model + prompt
+                if (
+                    "model" in kwargs
+                    and kwargs["model"] in local_cache[prompt]["models"]
+                ):
+                    cache_key = prompt + kwargs["model"]
+                    local_cache[cache_key] = result
+            else:  # caching based only on prompts
+                local_cache[prompt] = result
         except:
             pass
 
@@ -284,10 +301,9 @@ def client(original_function):
             function_setup(*args, **kwargs)
             ## MODEL CALL
             start_time = datetime.datetime.now()
-            if (
-                litellm.caching
-                and (cached_result := check_cache(*args, **kwargs)) is not None
-            ):
+            if (litellm.caching or litellm.caching_with_models) and (
+                cached_result := check_cache(*args, **kwargs)
+            ) is not None:
                 result = cached_result
             else:
                 result = original_function(*args, **kwargs)