Support for new gpt 4 turbo model

Author: alexeichhorn

tests/test_openai.py

@@ -26,7 +26,7 @@ def test_token_counter(self):
         ]
 
         # check if test covers all models (increase if new models are added)
-        assert len(OpenAIChatModel.__args__) == 14  #  type: ignore
+        assert len(OpenAIChatModel.__args__) == 16  #  type: ignore
 
         client = AsyncTypeOpenAI(api_key="mock")
 
@@ -42,9 +42,34 @@ def test_token_counter(self):
         assert client.chat.completions.num_tokens_from_messages(test_messages, model="gpt-4-32k") == 27
         assert client.chat.completions.num_tokens_from_messages(test_messages, model="gpt-4-32k-0314") == 27
         assert client.chat.completions.num_tokens_from_messages(test_messages, model="gpt-4-32k-0613") == 27
+        assert client.chat.completions.num_tokens_from_messages(test_messages, model="gpt-4-turbo-preview") == 27
         assert client.chat.completions.num_tokens_from_messages(test_messages, model="gpt-4-1106-preview") == 27
+        assert client.chat.completions.num_tokens_from_messages(test_messages, model="gpt-4-0125-preview") == 27
         assert client.chat.completions.num_tokens_from_messages(test_messages, model="gpt-4-vision-preview") == 27
 
+    def test_max_token_counter(self):
+        # check if test covers all models (increase if new models are added)
+        assert len(OpenAIChatModel.__args__) == 16  #  type: ignore
+
+        client = AsyncTypeOpenAI(api_key="mock")
+
+        assert client.chat.completions.max_tokens_of_model("gpt-3.5-turbo") == 4096
+        assert client.chat.completions.max_tokens_of_model("gpt-3.5-turbo-0301") == 4096
+        assert client.chat.completions.max_tokens_of_model("gpt-3.5-turbo-0613") == 4096
+        assert client.chat.completions.max_tokens_of_model("gpt-3.5-turbo-1106") == 16384
+        assert client.chat.completions.max_tokens_of_model("gpt-3.5-turbo-16k") == 16384
+        assert client.chat.completions.max_tokens_of_model("gpt-3.5-turbo-16k-0613") == 16384
+        assert client.chat.completions.max_tokens_of_model("gpt-4") == 8192
+        assert client.chat.completions.max_tokens_of_model("gpt-4-0314") == 8192
+        assert client.chat.completions.max_tokens_of_model("gpt-4-0613") == 8192
+        assert client.chat.completions.max_tokens_of_model("gpt-4-32k") == 32768
+        assert client.chat.completions.max_tokens_of_model("gpt-4-32k-0314") == 32768
+        assert client.chat.completions.max_tokens_of_model("gpt-4-32k-0613") == 32768
+        assert client.chat.completions.max_tokens_of_model("gpt-4-turbo-preview") == 128_000
+        assert client.chat.completions.max_tokens_of_model("gpt-4-1106-preview") == 128_000
+        assert client.chat.completions.max_tokens_of_model("gpt-4-0125-preview") == 128_000
+        assert client.chat.completions.max_tokens_of_model("gpt-4-vision-preview") == 128_000
+
     # -
 
     @pytest.fixture

typegpt/openai/base_chat_completion.py

@@ -15,7 +15,7 @@ def max_tokens_of_model(model: OpenAIChatModel) -> int:
                 return 8192
             case "gpt-4-32k" | "gpt-4-32k-0314" | "gpt-4-32k-0613":
                 return 32768
-            case "gpt-4-1106-preview" | "gpt-4-vision-preview":
+            case "gpt-4-turbo-preview" | "gpt-4-1106-preview" | "gpt-4-0125-preview" | "gpt-4-vision-preview":
                 return 128_000
 
     # copied from OpenAI: https://github.com/openai/openai-cookbook/blob/main/examples/How_to_count_tokens_with_tiktoken.ipynb
@@ -49,6 +49,8 @@ def num_tokens_from_messages(cls, messages: list[EncodedMessage], model: OpenAIC
             "gpt-3.5-turbo-16k-0613",
             "gpt-4-32k-0314",
             "gpt-4-32k-0613",
+            "gpt-4-turbo-preview",
+            "gpt-4-0125-preview",
             "gpt-4-1106-preview",
             "gpt-4-vision-preview",
         ):

typegpt/openai/views.py

@@ -14,7 +14,9 @@
     "gpt-4-32k",  # gpt-4 32k
     "gpt-4-32k-0314",
     "gpt-4-32k-0613",
-    "gpt-4-1106-preview",  # gpt-4 turbo
+    "gpt-4-turbo-preview",  # gpt-4 turbo
+    "gpt-4-1106-preview",
+    "gpt-4-0125-preview",
     "gpt-4-vision-preview",  # gpt-4 vision
 ]