openlit 1.33.11__py3-none-any.whl → 1.33.12__py3-none-any.whl

openlit/instrumentation/ai21/async_ai21.py

@@ -7,7 +7,6 @@ import time
 from opentelemetry.trace import SpanKind
 from openlit.__helpers import (
     handle_exception,
-    response_as_dict,
     set_server_address_and_port,
 )
 from openlit.instrumentation.ai21.utils import (
@@ -47,9 +46,9 @@ def async_chat(version, environment, application_name,
             self._span = span
             self._span_name = span_name
             # Placeholder for aggregating streaming response
-            self._llmresponse = ""
-            self._response_id = ""
-            self._finish_reason = ""
+            self._llmresponse = ''
+            self._response_id = ''
+            self._finish_reason = ''
             self._input_tokens = 0
             self._output_tokens = 0
             self._choices = []
@@ -100,7 +99,7 @@ def async_chat(version, environment, application_name,
                         )
                 except Exception as e:
                     handle_exception(self._span, e)
-                    logger.error("Error in trace creation: %s", e)
+                    logger.error('Error in trace creation: %s', e)
                 raise
 
     async def wrapper(wrapped, instance, args, kwargs):
@@ -109,12 +108,12 @@ def async_chat(version, environment, application_name,
         """
 
         # Check if streaming is enabled for the API call
-        streaming = kwargs.get("stream", False)
+        streaming = kwargs.get('stream', False)
 
-        server_address, server_port = set_server_address_and_port(instance, "api.ai21.com", 443)
-        request_model = kwargs.get("model", "jamba-1.5-mini")
+        server_address, server_port = set_server_address_and_port(instance, 'api.ai21.com', 443)
+        request_model = kwargs.get('model', 'jamba-1.5-mini')
 
-        span_name = f"{SemanticConvetion.GEN_AI_OPERATION_TYPE_CHAT} {request_model}"
+        span_name = f'{SemanticConvetion.GEN_AI_OPERATION_TYPE_CHAT} {request_model}'
 
         # pylint: disable=no-else-return
         if streaming:
@@ -129,7 +128,7 @@ def async_chat(version, environment, application_name,
                 start_time = time.time()
                 response = await wrapped(*args, **kwargs)
                 response = process_chat_response(
-                    response=response_as_dict(response),
+                    response=response,
                     request_model=request_model,
                     pricing_info=pricing_info,
                     server_port=server_port,
@@ -161,16 +160,16 @@ def async_chat_rag(version, environment, application_name,
         Wraps the GenAI function call.
         """
 
-        server_address, server_port = set_server_address_and_port(instance, "api.ai21.com", 443)
-        request_model = kwargs.get("model", "jamba-1.5-mini")
+        server_address, server_port = set_server_address_and_port(instance, 'api.ai21.com', 443)
+        request_model = kwargs.get('model', 'jamba-1.5-mini')
 
-        span_name = f"{SemanticConvetion.GEN_AI_OPERATION_TYPE_CHAT} {request_model}"
+        span_name = f'{SemanticConvetion.GEN_AI_OPERATION_TYPE_CHAT} {request_model}'
 
         with tracer.start_as_current_span(span_name, kind= SpanKind.CLIENT) as span:
             start_time = time.time()
             response = await wrapped(*args, **kwargs)
             response = process_chat_rag_response(
-                response=response_as_dict(response),
+                response=response,
                 request_model=request_model,
                 pricing_info=pricing_info,
                 server_port=server_port,

openlit/instrumentation/ai21/utils.py

@@ -29,30 +29,30 @@ def setup_common_span_attributes(span, request_model, kwargs, tokens,
     """
 
     # Base attributes from SDK and operation settings.
-    span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
+    span.set_attribute(TELEMETRY_SDK_NAME, 'openlit')
     span.set_attribute(SemanticConvetion.GEN_AI_OPERATION, SemanticConvetion.GEN_AI_OPERATION_TYPE_CHAT)
     span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM, SemanticConvetion.GEN_AI_SYSTEM_AI21)
     span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MODEL, request_model)
     span.set_attribute(SemanticConvetion.SERVER_PORT, server_port)
-    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_SEED, kwargs.get("seed", ""))
-    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_FREQUENCY_PENALTY, kwargs.get("frequency_penalty", 0.0))
-    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MAX_TOKENS, kwargs.get("max_tokens", -1))
-    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_PRESENCE_PENALTY, kwargs.get("presence_penalty", 0.0))
-    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_STOP_SEQUENCES, kwargs.get("stop", []))
-    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TEMPERATURE, kwargs.get("temperature", 0.4))
-    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_P, kwargs.get("top_p", 1.0))
+    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_SEED, kwargs.get('seed', ''))
+    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_FREQUENCY_PENALTY, kwargs.get('frequency_penalty', 0.0))
+    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MAX_TOKENS, kwargs.get('max_tokens', -1))
+    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_PRESENCE_PENALTY, kwargs.get('presence_penalty', 0.0))
+    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_STOP_SEQUENCES, kwargs.get('stop', []))
+    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TEMPERATURE, kwargs.get('temperature', 0.4))
+    span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_P, kwargs.get('top_p', 1.0))
 
     # Add token-related attributes if available.
-    if "finish_reason" in tokens:
-        span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_FINISH_REASON, [tokens["finish_reason"]])
-    if "response_id" in tokens:
-        span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_ID, tokens["response_id"])
-    if "input_tokens" in tokens:
-        span.set_attribute(SemanticConvetion.GEN_AI_USAGE_INPUT_TOKENS, tokens["input_tokens"])
-    if "output_tokens" in tokens:
-        span.set_attribute(SemanticConvetion.GEN_AI_USAGE_OUTPUT_TOKENS, tokens["output_tokens"])
-    if "total_tokens" in tokens:
-        span.set_attribute(SemanticConvetion.GEN_AI_USAGE_TOTAL_TOKENS, tokens["total_tokens"])
+    if 'finish_reason' in tokens:
+        span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_FINISH_REASON, [tokens['finish_reason']])
+    if 'response_id' in tokens:
+        span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_ID, tokens['response_id'])
+    if 'input_tokens' in tokens:
+        span.set_attribute(SemanticConvetion.GEN_AI_USAGE_INPUT_TOKENS, tokens['input_tokens'])
+    if 'output_tokens' in tokens:
+        span.set_attribute(SemanticConvetion.GEN_AI_USAGE_OUTPUT_TOKENS, tokens['output_tokens'])
+    if 'total_tokens' in tokens:
+        span.set_attribute(SemanticConvetion.GEN_AI_USAGE_TOTAL_TOKENS, tokens['total_tokens'])
 
     span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_MODEL, request_model)
     span.set_attribute(SemanticConvetion.SERVER_ADDRESS, server_address)
@@ -80,15 +80,15 @@ def record_common_metrics(metrics, application_name, environment, request_model,
         server_port=server_port,
         response_model=request_model,
     )
-    metrics["genai_client_usage_tokens"].record(input_tokens + output_tokens, attributes)
-    metrics["genai_client_operation_duration"].record(end_time - start_time, attributes)
+    metrics['genai_client_usage_tokens'].record(input_tokens + output_tokens, attributes)
+    metrics['genai_client_operation_duration'].record(end_time - start_time, attributes)
     if include_tbt and tbt_value is not None:
-        metrics["genai_server_tbt"].record(tbt_value, attributes)
-    metrics["genai_server_ttft"].record(end_time - start_time, attributes)
-    metrics["genai_requests"].add(1, attributes)
-    metrics["genai_completion_tokens"].add(output_tokens, attributes)
-    metrics["genai_prompt_tokens"].add(input_tokens, attributes)
-    metrics["genai_cost"].record(cost, attributes)
+        metrics['genai_server_tbt'].record(tbt_value, attributes)
+    metrics['genai_server_ttft'].record(end_time - start_time, attributes)
+    metrics['genai_requests'].add(1, attributes)
+    metrics['genai_completion_tokens'].add(output_tokens, attributes)
+    metrics['genai_prompt_tokens'].add(input_tokens, attributes)
+    metrics['genai_cost'].record(cost, attributes)
 
 def emit_common_events(event_provider, choices, finish_reason, llmresponse, formatted_messages,
                        capture_message_content, n):
@@ -99,25 +99,25 @@ def emit_common_events(event_provider, choices, finish_reason, llmresponse, form
     if n > 1:
         for choice in choices:
             choice_event_body = {
-                "finish_reason": finish_reason,
-                "index": choice.get('index', 0),
-                "message": {
-                    **({"content": choice.get('message', {}).get('content', '')} if capture_message_content else {}),
-                    "role": choice.get('message', {}).get('role', 'assistant')
+                'finish_reason': finish_reason,
+                'index': choice.get('index', 0),
+                'message': {
+                    **({'content': choice.get('message', {}).get('content', '')} if capture_message_content else {}),
+                    'role': choice.get('message', {}).get('role', 'assistant')
                 }
             }
             # If tool calls exist, emit an event for each tool call.
             tool_calls = choice.get('message', {}).get('tool_calls')
             if tool_calls:
                 for tool_call in tool_calls:
-                    choice_event_body["message"].update({
-                        "tool_calls": {
-                            "function": {
-                                "name": tool_call.get('function', {}).get('name', ''),
-                                "arguments": tool_call.get('function', {}).get('arguments', '')
+                    choice_event_body['message'].update({
+                        'tool_calls': {
+                            'function': {
+                                'name': tool_call.get('function', {}).get('name', ''),
+                                'arguments': tool_call.get('function', {}).get('arguments', '')
                             },
-                            "id": tool_call.get('id', ''),
-                            "type": tool_call.get('type', 'function')
+                            'id': tool_call.get('id', ''),
+                            'type': tool_call.get('type', 'function')
                         }
                     })
                     event = otel_event(
@@ -136,11 +136,11 @@ def emit_common_events(event_provider, choices, finish_reason, llmresponse, form
     else:
         # Single choice case.
         choice_event_body = {
-            "finish_reason": finish_reason,
-            "index": 0,
-            "message": {
-                **({"content": llmresponse} if capture_message_content else {}),
-                "role": 'assistant'
+            'finish_reason': finish_reason,
+            'index': 0,
+            'message': {
+                **({'content': llmresponse} if capture_message_content else {}),
+                'role': 'assistant'
             }
         }
         event = otel_event(
@@ -155,25 +155,25 @@ def emit_common_events(event_provider, choices, finish_reason, llmresponse, form
         msg = formatted_messages.get(role, {})
         if msg.get('content', ''):
             event_body = {
-                **({"content": msg.get('content', '')} if capture_message_content else {}),
-                "role": msg.get('role', [])
+                **({'content': msg.get('content', '')} if capture_message_content else {}),
+                'role': msg.get('role', [])
             }
             # For assistant messages, attach tool call details if they exist.
             if role == 'assistant' and choices:
                 tool_calls = choices[0].get('message', {}).get('tool_calls', [])
                 if tool_calls:
-                    event_body["tool_calls"] = {
-                        "function": {
-                            "name": tool_calls[0].get('function', {}).get('name', ''),
-                            "arguments": tool_calls[0].get('function', {}).get('arguments', '')
+                    event_body['tool_calls'] = {
+                        'function': {
+                            'name': tool_calls[0].get('function', {}).get('name', ''),
+                            'arguments': tool_calls[0].get('function', {}).get('arguments', '')
                         },
-                        "id": tool_calls[0].get('id', ''),
-                        "type": "function"
+                        'id': tool_calls[0].get('id', ''),
+                        'type': 'function'
                     }
             if role == 'tool' and choices:
                 tool_calls = choices[0].get('message', {}).get('tool_calls', [])
                 if tool_calls:
-                    event_body["id"] = tool_calls[0].get('id', '')
+                    event_body['id'] = tool_calls[0].get('id', '')
             event = otel_event(
                 name=getattr(SemanticConvetion, f'GEN_AI_{role.upper()}_MESSAGE'),
                 attributes={SemanticConvetion.GEN_AI_SYSTEM: SemanticConvetion.GEN_AI_SYSTEM_AI21},
@@ -201,8 +201,8 @@ def process_chunk(self, chunk):
         if content:
             self._llmresponse += content
         if chunked.get('usage'):
-            self._input_tokens = chunked.get('usage').get("prompt_tokens")
-            self._output_tokens = chunked.get('usage').get("completion_tokens")
+            self._input_tokens = chunked.get('usage').get('prompt_tokens')
+            self._output_tokens = chunked.get('usage').get('completion_tokens')
     self._response_id = chunked.get('id')
     self._choices += chunked.get('choices')
     self._finish_reason = chunked.get('choices')[0].get('finish_reason')
@@ -218,19 +218,19 @@ def common_chat_logic(scope, pricing_info, environment, application_name, metric
         scope._tbt = calculate_tbt(scope._timestamps)
 
     # Extract and format input messages.
-    formatted_messages = extract_and_format_input(scope._kwargs.get("messages", ""))
+    formatted_messages = extract_and_format_input(scope._kwargs.get('messages', ''))
     prompt = concatenate_all_contents(formatted_messages)
-    request_model = scope._kwargs.get("model", "jamba-1.5-mini")
+    request_model = scope._kwargs.get('model', 'jamba-1.5-mini')
 
     # Calculate cost based on token usage.
     cost = get_chat_model_cost(request_model, pricing_info, scope._input_tokens, scope._output_tokens)
     # Prepare tokens dictionary.
     tokens = {
-        "finish_reason": scope._finish_reason,
-        "response_id": scope._response_id,
-        "input_tokens": scope._input_tokens,
-        "output_tokens": scope._output_tokens,
-        "total_tokens": scope._input_tokens + scope._output_tokens,
+        'finish_reason': scope._finish_reason,
+        'response_id': scope._response_id,
+        'input_tokens': scope._input_tokens,
+        'output_tokens': scope._output_tokens,
+        'total_tokens': scope._input_tokens + scope._output_tokens,
     }
     extra_attrs = {
         SemanticConvetion.GEN_AI_REQUEST_IS_STREAM: is_stream,
@@ -239,7 +239,7 @@ def common_chat_logic(scope, pricing_info, environment, application_name, metric
         SemanticConvetion.GEN_AI_SERVER_TBT: scope._tbt,
         SemanticConvetion.GEN_AI_SERVER_TTFT: scope._ttft,
         SemanticConvetion.GEN_AI_SDK_VERSION: version,
-        SemanticConvetion.GEN_AI_OUTPUT_TYPE: "text" if isinstance(scope._llmresponse, str) else "json"
+        SemanticConvetion.GEN_AI_OUTPUT_TYPE: 'text' if isinstance(scope._llmresponse, str) else 'json'
     }
     # Set span attributes.
     setup_common_span_attributes(scope._span, request_model, scope._kwargs, tokens,
@@ -282,13 +282,14 @@ def process_streaming_chat_response(self, pricing_info, environment, application
 
 def process_chat_response(response, request_model, pricing_info, server_port, server_address,
                           environment, application_name, metrics, event_provider, start_time,
-                          span, capture_message_content=False, disable_metrics=False, version="1.0.0", **kwargs):
+                          span, capture_message_content=False, disable_metrics=False, version='1.0.0', **kwargs):
     """
     Process a synchronous chat response and generate Telemetry.
     """
 
     # Create a generic scope object to hold telemetry data.
     self = type('GenericScope', (), {})()
+    response_dict = response_as_dict(response)
 
     # pylint: disable = no-member
     self._start_time = start_time
@@ -298,19 +299,19 @@ def process_chat_response(response, request_model, pricing_info, server_port, se
     # Concatenate content from all choices.
     self._llmresponse = ''.join(
         (choice.get('message', {}).get('content') or '')
-        for choice in response.get('choices', [])
+        for choice in response_dict.get('choices', [])
     )
-    self._response_role = response.get('message', {}).get('role', 'assistant')
-    self._input_tokens = response.get('usage', {}).get('prompt_tokens', 0)
-    self._output_tokens = response.get('usage', {}).get('completion_tokens', 0)
-    self._response_id = response.get('id', '')
+    self._response_role = response_dict.get('message', {}).get('role', 'assistant')
+    self._input_tokens = response_dict.get('usage', {}).get('prompt_tokens', 0)
+    self._output_tokens = response_dict.get('usage', {}).get('completion_tokens', 0)
+    self._response_id = response_dict.get('id', '')
     self._response_model = request_model
-    self._finish_reason = response.get('choices', [{}])[0].get('finish_reason')
+    self._finish_reason = response_dict.get('choices', [{}])[0].get('finish_reason')
     self._timestamps = []
     self._ttft, self._tbt = self._end_time - self._start_time, 0
     self._server_address, self._server_port = server_address, server_port
     self._kwargs = kwargs
-    self._choices = response.get('choices')
+    self._choices = response_dict.get('choices')
 
     common_chat_logic(self, pricing_info, environment, application_name, metrics,
                       event_provider, capture_message_content, disable_metrics, version, is_stream=False)
@@ -319,30 +320,31 @@ def process_chat_response(response, request_model, pricing_info, server_port, se
 
 def process_chat_rag_response(response, request_model, pricing_info, server_port, server_address,
                               environment, application_name, metrics, event_provider, start_time,
-                              span, capture_message_content=False, disable_metrics=False, version="1.0.0", **kwargs):
+                              span, capture_message_content=False, disable_metrics=False, version='1.0.0', **kwargs):
     """
     Process a chat response and generate Telemetry.
     """
     end_time = time.time()
+    response_dict = response_as_dict(response)
     try:
         # Format input messages into a single prompt string.
-        messages_input = kwargs.get("messages", "")
+        messages_input = kwargs.get('messages', '')
         formatted_messages = extract_and_format_input(messages_input)
         prompt = concatenate_all_contents(formatted_messages)
         input_tokens = general_tokens(prompt)
 
         # Create tokens dict and RAG-specific extra attributes.
-        tokens = {"response_id": response.get("id"), "input_tokens": input_tokens}
+        tokens = {'response_id': response_dict.get('id'), 'input_tokens': input_tokens}
         extra_attrs = {
             SemanticConvetion.GEN_AI_REQUEST_IS_STREAM: False,
             SemanticConvetion.GEN_AI_SERVER_TTFT: end_time - start_time,
             SemanticConvetion.GEN_AI_SDK_VERSION: version,
-            SemanticConvetion.GEN_AI_RAG_MAX_SEGMENTS: kwargs.get("max_segments", -1),
-            SemanticConvetion.GEN_AI_RAG_STRATEGY: kwargs.get("retrieval_strategy", "segments"),
-            SemanticConvetion.GEN_AI_RAG_SIMILARITY_THRESHOLD: kwargs.get("retrieval_similarity_threshold", -1),
-            SemanticConvetion.GEN_AI_RAG_MAX_NEIGHBORS: kwargs.get("max_neighbors", -1),
-            SemanticConvetion.GEN_AI_RAG_FILE_IDS: str(kwargs.get("file_ids", "")),
-            SemanticConvetion.GEN_AI_RAG_DOCUMENTS_PATH: kwargs.get("path", "")
+            SemanticConvetion.GEN_AI_RAG_MAX_SEGMENTS: kwargs.get('max_segments', -1),
+            SemanticConvetion.GEN_AI_RAG_STRATEGY: kwargs.get('retrieval_strategy', 'segments'),
+            SemanticConvetion.GEN_AI_RAG_SIMILARITY_THRESHOLD: kwargs.get('retrieval_similarity_threshold', -1),
+            SemanticConvetion.GEN_AI_RAG_MAX_NEIGHBORS: kwargs.get('max_neighbors', -1),
+            SemanticConvetion.GEN_AI_RAG_FILE_IDS: str(kwargs.get('file_ids', '')),
+            SemanticConvetion.GEN_AI_RAG_DOCUMENTS_PATH: kwargs.get('path', '')
         }
         # Set common span attributes.
         setup_common_span_attributes(span, request_model, kwargs, tokens,
@@ -357,7 +359,7 @@ def process_chat_rag_response(response, request_model, pricing_info, server_port
             )
 
         output_tokens = 0
-        choices = response.get('choices', [])
+        choices = response_dict.get('choices', [])
         # Instead of adding a separate event per choice, we aggregate all completion content.
         aggregated_completion = []
         for i in range(kwargs.get('n', 1)):
@@ -370,14 +372,14 @@ def process_chat_rag_response(response, request_model, pricing_info, server_port
                                    str(choices[i].get('message', {}).get('tool_calls')))
             # Set output type based on actual content type.
             if isinstance(content, str):
-                span.set_attribute(SemanticConvetion.GEN_AI_OUTPUT_TYPE, "text")
+                span.set_attribute(SemanticConvetion.GEN_AI_OUTPUT_TYPE, 'text')
             elif content is not None:
-                span.set_attribute(SemanticConvetion.GEN_AI_OUTPUT_TYPE, "json")
+                span.set_attribute(SemanticConvetion.GEN_AI_OUTPUT_TYPE, 'json')
 
         # Concatenate completion responses.
         llmresponse = ''.join(aggregated_completion)
-        tokens["output_tokens"] = output_tokens
-        tokens["total_tokens"] = input_tokens + output_tokens
+        tokens['output_tokens'] = output_tokens
+        tokens['total_tokens'] = input_tokens + output_tokens
 
         cost = get_chat_model_cost(request_model, pricing_info, input_tokens, output_tokens)
         span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COST, cost)

openlit/instrumentation/anthropic/__init__.py

@@ -1,4 +1,3 @@
-# pylint: disable=useless-return, bad-staticmethod-argument, disable=duplicate-code
 """Initializer of Auto Instrumentation of Anthropic Functions"""
 
 from typing import Collection
@@ -9,7 +8,7 @@ from wrapt import wrap_function_wrapper
 from openlit.instrumentation.anthropic.anthropic import messages
 from openlit.instrumentation.anthropic.async_anthropic import async_messages
 
-_instruments = ("anthropic >= 0.21.0",)
+_instruments = ('anthropic >= 0.21.0',)
 
 class AnthropicInstrumentor(BaseInstrumentor):
     """
@@ -20,29 +19,30 @@ class AnthropicInstrumentor(BaseInstrumentor):
         return _instruments
 
     def _instrument(self, **kwargs):
-        application_name = kwargs.get("application_name", "default")
-        environment = kwargs.get("environment", "default")
-        tracer = kwargs.get("tracer")
-        metrics = kwargs.get("metrics_dict")
-        pricing_info = kwargs.get("pricing_info", {})
-        capture_message_content = kwargs.get("capture_message_content", False)
-        disable_metrics = kwargs.get("disable_metrics")
-        version = importlib.metadata.version("anthropic")
+        application_name = kwargs.get('application_name', 'default')
+        environment = kwargs.get('environment', 'default')
+        tracer = kwargs.get('tracer')
+        event_provider = kwargs.get('event_provider')
+        metrics = kwargs.get('metrics_dict')
+        pricing_info = kwargs.get('pricing_info', {})
+        capture_message_content = kwargs.get('capture_message_content', False)
+        disable_metrics = kwargs.get('disable_metrics')
+        version = importlib.metadata.version('anthropic')
 
         #sync
         wrap_function_wrapper(
-            "anthropic.resources.messages",  
-            "Messages.create",  
+            'anthropic.resources.messages',  
+            'Messages.create',  
             messages(version, environment, application_name,
-                     tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+                     tracer, event_provider, pricing_info, capture_message_content, metrics, disable_metrics),
         )
 
         #async
         wrap_function_wrapper(
-            "anthropic.resources.messages",  
-            "AsyncMessages.create",  
+            'anthropic.resources.messages',  
+            'AsyncMessages.create',  
             async_messages(version, environment, application_name,
-                            tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+                            tracer, event_provider, pricing_info, capture_message_content, metrics, disable_metrics),
         )
 
     def _uninstrument(self, **kwargs):