vectorvein 0.2.78__tar.gz → 0.2.80__tar.gz

{vectorvein-0.2.78 → vectorvein-0.2.80}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.2.78
+Version: 0.2.80
 Summary: VectorVein Python SDK
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.2.78 → vectorvein-0.2.80}/pyproject.toml

@@ -17,7 +17,7 @@ description = "VectorVein Python SDK"
 name = "vectorvein"
 readme = "README.md"
 requires-python = ">=3.10"
-version = "0.2.78"
+version = "0.2.80"
 
 [project.license]
 text = "MIT"

{vectorvein-0.2.78 → vectorvein-0.2.80}/src/vectorvein/chat_clients/anthropic_client.py

@@ -596,7 +596,10 @@ class AnthropicChatClient(BaseChatClient):
 
         if max_tokens is None:
             max_output_tokens = self.model_setting.max_output_tokens
-            token_counts = get_message_token_counts(messages=messages, tools=tools_params, model=self.model)
+            native_multimodal = self.model_setting.native_multimodal
+            token_counts = get_message_token_counts(
+                messages=messages, tools=tools, model=self.model, native_multimodal=native_multimodal
+            )
             if max_output_tokens is not None:
                 max_tokens = self.model_setting.context_length - token_counts
                 max_tokens = min(max(max_tokens, 1), max_output_tokens)
@@ -1207,7 +1210,10 @@ class AsyncAnthropicChatClient(BaseAsyncChatClient):
 
         if max_tokens is None:
             max_output_tokens = self.model_setting.max_output_tokens
-            token_counts = get_message_token_counts(messages=messages, tools=tools_params, model=self.model)
+            native_multimodal = self.model_setting.native_multimodal
+            token_counts = get_message_token_counts(
+                messages=messages, tools=tools, model=self.model, native_multimodal=native_multimodal
+            )
             if max_output_tokens is not None:
                 max_tokens = self.model_setting.context_length - token_counts
                 max_tokens = min(max(max_tokens, 1), max_output_tokens)

{vectorvein-0.2.78 → vectorvein-0.2.80}/src/vectorvein/chat_clients/openai_compatible_client.py

@@ -326,7 +326,10 @@ class OpenAICompatibleChatClient(BaseChatClient):
 
         if not max_tokens and not max_completion_tokens:
             max_output_tokens = self.model_setting.max_output_tokens
-            token_counts = get_message_token_counts(messages=messages, tools=tools, model=self.model)
+            native_multimodal = self.model_setting.native_multimodal
+            token_counts = get_message_token_counts(
+                messages=messages, tools=tools, model=self.model, native_multimodal=native_multimodal
+            )
             if max_output_tokens is not None:
                 max_tokens = self.model_setting.context_length - token_counts - 64
                 max_tokens = min(max(max_tokens, 1), max_output_tokens)
@@ -805,7 +808,10 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
 
         if not max_tokens and not max_completion_tokens:
             max_output_tokens = self.model_setting.max_output_tokens
-            token_counts = get_message_token_counts(messages=messages, tools=tools, model=self.model)
+            native_multimodal = self.model_setting.native_multimodal
+            token_counts = get_message_token_counts(
+                messages=messages, tools=tools, model=self.model, native_multimodal=native_multimodal
+            )
             if max_output_tokens is not None:
                 max_tokens = self.model_setting.context_length - token_counts - 64
                 max_tokens = min(max(max_tokens, 1), max_output_tokens)

{vectorvein-0.2.78 → vectorvein-0.2.80}/src/vectorvein/chat_clients/utils.py

@@ -348,9 +348,10 @@ def get_message_token_counts(
     messages: list,
     tools: list | Iterable[ToolParam] | NotGiven = NOT_GIVEN,
     model: str = "gpt-4o",
+    native_multimodal: bool = True,
 ) -> int:
     tokens = 0
-    formatted_messages = format_messages(messages, backend=BackendType.OpenAI, native_multimodal=True)
+    formatted_messages = format_messages(messages, backend=BackendType.OpenAI, native_multimodal=native_multimodal)
     for message in formatted_messages:
         content = message["content"]
         if isinstance(content, str):
@@ -360,8 +361,11 @@ def get_message_token_counts(
                 if isinstance(item, dict) and item["type"] == "text":
                     tokens += get_token_counts(item["text"], model)
                 elif isinstance(item, dict) and item["type"].startswith("image"):
-                    # TODO: Get real image size
-                    tokens += calculate_image_tokens(2048, 2048, model)
+                    if native_multimodal:
+                        # TODO: Get real image size
+                        tokens += calculate_image_tokens(2048, 2048, model)
+                    else:
+                        tokens += 1
     if tools:
         tokens += get_token_counts(str(tools), model)
 

{vectorvein-0.2.78 → vectorvein-0.2.80}/src/vectorvein/types/defaults.py

@@ -607,6 +607,14 @@ OPENAI_MODELS: Final[Dict[str, ModelSettingDict]] = {
         "response_format_available": True,
         "native_multimodal": False,
     },
+    "gpt-4.1": {
+        "id": "gpt-4.1",
+        "context_length": 1047576,
+        "max_output_tokens": 32768,
+        "function_call_available": True,
+        "response_format_available": True,
+        "native_multimodal": True,
+    },
 }
 
 # Anthropic models