openlit 1.17.0__py3-none-any.whl → 1.18.1__py3-none-any.whl

openlit/__init__.py

@@ -32,6 +32,7 @@ from openlit.instrumentation.groq import GroqInstrumentor
 from openlit.instrumentation.ollama import OllamaInstrumentor
 from openlit.instrumentation.gpt4all import GPT4AllInstrumentor
 from openlit.instrumentation.elevenlabs import ElevenLabsInstrumentor
+from openlit.instrumentation.vllm import VLLMInstrumentor
 from openlit.instrumentation.langchain import LangChainInstrumentor
 from openlit.instrumentation.llamaindex import LlamaIndexInstrumentor
 from openlit.instrumentation.haystack import HaystackInstrumentor
@@ -194,6 +195,7 @@ def init(environment="default", application_name="default", tracer=None, otlp_en
         "ollama": "ollama",
         "gpt4all": "gpt4all",
         "elevenlabs": "elevenlabs",
+        "vllm": "vllm",
         "langchain": "langchain",
         "llama_index": "llama_index",
         "haystack": "haystack",
@@ -270,6 +272,7 @@ def init(environment="default", application_name="default", tracer=None, otlp_en
             "ollama": OllamaInstrumentor(),
             "gpt4all": GPT4AllInstrumentor(),
             "elevenlabs": ElevenLabsInstrumentor(),
+            "vllm": VLLMInstrumentor(),
             "langchain": LangChainInstrumentor(),
             "llama_index": LlamaIndexInstrumentor(),
             "haystack": HaystackInstrumentor(),

openlit/instrumentation/ollama/async_ollama.py

@@ -6,7 +6,11 @@ Module for monitoring Ollama API calls.
 import logging
 from opentelemetry.trace import SpanKind, Status, StatusCode
 from opentelemetry.sdk.resources import TELEMETRY_SDK_NAME
-from openlit.__helpers import handle_exception, general_tokens
+from openlit.__helpers import (
+    handle_exception,
+    general_tokens,
+    get_chat_model_cost,
+    get_embed_model_cost)
 from openlit.semcov import SemanticConvetion
 
 # Initialize logger for logging potential issues and operations
@@ -90,10 +94,11 @@ def async_chat(gen_ai_endpoint, version, environment, application_name,
                                 formatted_messages.append(f"{role}: {content}")
                         prompt = "\n".join(formatted_messages)
 
-                        # Calculate cost of the operation
-                        cost = 0
                         prompt_tokens = general_tokens(prompt)
                         total_tokens = prompt_tokens + completion_tokens
+                        # Calculate cost of the operation
+                        cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                         # Set Span attributes
                         span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
@@ -219,11 +224,12 @@ def async_chat(gen_ai_endpoint, version, environment, application_name,
                             },
                         )
 
-                    # Calculate cost of the operation
-                    cost = 0
                     prompt_tokens = general_tokens(prompt)
                     completion_tokens = response["eval_count"]
                     total_tokens = prompt_tokens + completion_tokens
+                    # Calculate cost of the operation
+                    cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                     span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
                                         prompt_tokens)
@@ -331,10 +337,11 @@ def async_generate(gen_ai_endpoint, version, environment, application_name,
 
                     # Handling exception ensure observability without disrupting operation
                     try:
-                        # Calculate cost of the operation
-                        cost = 0
                         prompt_tokens = general_tokens(kwargs.get("prompt", ""))
                         total_tokens = prompt_tokens + completion_tokens
+                        # Calculate cost of the operation
+                        cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                         # Set Span attributes
                         span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
@@ -442,11 +449,12 @@ def async_generate(gen_ai_endpoint, version, environment, application_name,
                             },
                         )
 
-                    # Calculate cost of the operation
-                    cost = 0
                     prompt_tokens = response["prompt_eval_count"]
                     completion_tokens = response["eval_count"]
                     total_tokens = prompt_tokens + completion_tokens
+                    # Calculate cost of the operation
+                    cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                     span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
                                         prompt_tokens)
@@ -534,9 +542,10 @@ def async_embeddings(gen_ai_endpoint, version, environment, application_name,
             response = await wrapped(*args, **kwargs)
 
             try:
-                # Calculate cost of the operation
-                cost = 0
                 prompt_tokens = general_tokens(kwargs.get('prompt', ""))
+                # Calculate cost of the operation
+                cost = get_embed_model_cost(kwargs.get('model', "mistral-embed"),
+                                            pricing_info, prompt_tokens)
                 # Set Span attributes
                 span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
                 span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,

openlit/instrumentation/ollama/ollama.py

@@ -6,7 +6,12 @@ Module for monitoring Ollama API calls.
 import logging
 from opentelemetry.trace import SpanKind, Status, StatusCode
 from opentelemetry.sdk.resources import TELEMETRY_SDK_NAME
-from openlit.__helpers import handle_exception, general_tokens
+from openlit.__helpers import (
+    handle_exception,
+    general_tokens,
+    get_chat_model_cost,
+    get_embed_model_cost
+)
 from openlit.semcov import SemanticConvetion
 
 # Initialize logger for logging potential issues and operations
@@ -90,10 +95,11 @@ def chat(gen_ai_endpoint, version, environment, application_name,
                                 formatted_messages.append(f"{role}: {content}")
                         prompt = "\n".join(formatted_messages)
 
-                        # Calculate cost of the operation
-                        cost = 0
                         prompt_tokens = general_tokens(prompt)
                         total_tokens = prompt_tokens + completion_tokens
+                        # Calculate cost of the operation
+                        cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                         # Set Span attributes
                         span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
@@ -219,11 +225,12 @@ def chat(gen_ai_endpoint, version, environment, application_name,
                             },
                         )
 
-                    # Calculate cost of the operation
-                    cost = 0
                     prompt_tokens = general_tokens(prompt)
                     completion_tokens = response["eval_count"]
                     total_tokens = prompt_tokens + completion_tokens
+                    # Calculate cost of the operation
+                    cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                     span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
                                         prompt_tokens)
@@ -331,10 +338,11 @@ def generate(gen_ai_endpoint, version, environment, application_name,
 
                     # Handling exception ensure observability without disrupting operation
                     try:
-                        # Calculate cost of the operation
-                        cost = 0
                         prompt_tokens = general_tokens(kwargs.get("prompt", ""))
                         total_tokens = prompt_tokens + completion_tokens
+                        # Calculate cost of the operation
+                        cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                         # Set Span attributes
                         span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
@@ -442,11 +450,12 @@ def generate(gen_ai_endpoint, version, environment, application_name,
                             },
                         )
 
-                    # Calculate cost of the operation
-                    cost = 0
                     prompt_tokens = response["prompt_eval_count"]
                     completion_tokens = response["eval_count"]
                     total_tokens = prompt_tokens + completion_tokens
+                    # Calculate cost of the operation
+                    cost = get_chat_model_cost(kwargs.get("model", "llama3"),
+                                                pricing_info, prompt_tokens, completion_tokens)
 
                     span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
                                         prompt_tokens)
@@ -534,9 +543,10 @@ def embeddings(gen_ai_endpoint, version, environment, application_name,
             response = wrapped(*args, **kwargs)
 
             try:
-                # Calculate cost of the operation
-                cost = 0
                 prompt_tokens = general_tokens(kwargs.get('prompt', ""))
+                # Calculate cost of the operation
+                cost = get_embed_model_cost(kwargs.get('model', "mistral-embed"),
+                                            pricing_info, prompt_tokens)
                 # Set Span attributes
                 span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
                 span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,

openlit/instrumentation/vllm/__init__.py

@@ -0,0 +1,43 @@
+# pylint: disable=useless-return, bad-staticmethod-argument, disable=duplicate-code
+"""Initializer of Auto Instrumentation of vLLM Functions"""
+
+from typing import Collection
+import importlib.metadata
+from opentelemetry.instrumentation.instrumentor import BaseInstrumentor
+from wrapt import wrap_function_wrapper
+
+from openlit.instrumentation.vllm.vllm import (
+    generate
+)
+
+_instruments = ("vllm >= 0.5.4",)
+
+class VLLMInstrumentor(BaseInstrumentor):
+    """
+    An instrumentor for vLLM's client library.
+    """
+
+    def instrumentation_dependencies(self) -> Collection[str]:
+        return _instruments
+
+    def _instrument(self, **kwargs):
+        application_name = kwargs.get("application_name", "default_application")
+        environment = kwargs.get("environment", "default_environment")
+        tracer = kwargs.get("tracer")
+        metrics = kwargs.get("metrics_dict")
+        pricing_info = kwargs.get("pricing_info", {})
+        trace_content = kwargs.get("trace_content", False)
+        disable_metrics = kwargs.get("disable_metrics")
+        version = importlib.metadata.version("vllm")
+
+        # sync chat
+        wrap_function_wrapper(
+            "vllm",
+            "LLM.generate",
+            generate("vllm.generate", version, environment, application_name,
+                  tracer, pricing_info, trace_content, metrics, disable_metrics),
+        )
+
+    def _uninstrument(self, **kwargs):
+        # Proper uninstrumentation logic to revert patched methods
+        pass

openlit/instrumentation/vllm/vllm.py

@@ -0,0 +1,143 @@
+# pylint: disable=duplicate-code, broad-exception-caught, too-many-statements, unused-argument, possibly-used-before-assignment
+"""
+Module for monitoring vLLM API calls.
+"""
+
+import logging
+from opentelemetry.trace import SpanKind, Status, StatusCode
+from opentelemetry.sdk.resources import TELEMETRY_SDK_NAME
+from openlit.__helpers import handle_exception, general_tokens
+from openlit.semcov import SemanticConvetion
+
+# Initialize logger for logging potential issues and operations
+logger = logging.getLogger(__name__)
+
+def generate(gen_ai_endpoint, version, environment, application_name,
+                     tracer, pricing_info, trace_content, metrics, disable_metrics):
+    """
+    Generates a telemetry wrapper for generate to collect metrics.
+
+    Args:
+        gen_ai_endpoint: Endpoint identifier for logging and tracing.
+        version: Version of the monitoring package.
+        environment: Deployment environment (e.g., production, staging).
+        application_name: Name of the application using the vLLM API.
+        tracer: OpenTelemetry tracer for creating spans.
+        pricing_info: Information used for calculating the cost of vLLM usage.
+        trace_content: Flag indicating whether to trace the actual content.
+
+    Returns:
+        A function that wraps the generate method to add telemetry.
+    """
+
+    def wrapper(wrapped, instance, args, kwargs):
+        """
+        Wraps the 'generate' API call to add telemetry.
+        
+        This collects metrics such as execution time, cost, and token usage, and handles errors
+        gracefully, adding details to the trace for observability.
+
+        Args:
+            wrapped: The original 'generate' method to be wrapped.
+            instance: The instance of the class where the original method is defined.
+            args: Positional arguments for the 'generate' method.
+            kwargs: Keyword arguments for the 'generate' method.
+
+        Returns:
+            The response from the original 'generate' method.
+        """
+
+        # pylint: disable=line-too-long
+        with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
+            response = wrapped(*args, **kwargs)
+
+            try:
+                model = instance.llm_engine.model_config.model or "facebook/opt-125m"
+                # Set base span attribues
+                span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
+                span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
+                                    SemanticConvetion.GEN_AI_SYSTEM_VLLM)
+                span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
+                                    SemanticConvetion.GEN_AI_TYPE_CHAT)
+                span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
+                                    gen_ai_endpoint)
+                span.set_attribute(SemanticConvetion.GEN_AI_ENVIRONMENT,
+                                    environment)
+                span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
+                                    application_name)
+                span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MODEL,
+                                    model)
+                span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_IS_STREAM,
+                                    False)
+                input_tokens = 0
+                output_tokens = 0
+                cost = 0
+
+                if trace_content:
+                    prompt_attributes = {}
+                    completion_attributes = {}
+
+                    for i, output in enumerate(response):
+                        prompt_attributes[f"{SemanticConvetion.GEN_AI_CONTENT_PROMPT}.{i}"] = output.prompt
+                        completion_attributes[f"{SemanticConvetion.GEN_AI_CONTENT_COMPLETION}.{i}"] = output.outputs[0].text
+                        input_tokens += general_tokens(output.prompt)
+                        output_tokens += general_tokens(output.outputs[0].text)
+
+                    # Add a single event for all prompts
+                    span.add_event(
+                        name=SemanticConvetion.GEN_AI_CONTENT_PROMPT_EVENT,
+                        attributes=prompt_attributes,
+                    )
+
+                    # Add a single event for all completions
+                    span.add_event(
+                        name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
+                        attributes=completion_attributes,
+                    )
+
+                total_tokens = input_tokens + output_tokens
+
+                span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
+                                    input_tokens)
+                span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COMPLETION_TOKENS,
+                                    output_tokens)
+                span.set_attribute(SemanticConvetion.GEN_AI_USAGE_TOTAL_TOKENS,
+                                    total_tokens)
+                span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COST,
+                                    cost)
+
+                span.set_status(Status(StatusCode.OK))
+
+                if disable_metrics is False:
+                    attributes = {
+                        TELEMETRY_SDK_NAME:
+                            "openlit",
+                        SemanticConvetion.GEN_AI_APPLICATION_NAME:
+                            application_name,
+                        SemanticConvetion.GEN_AI_SYSTEM:
+                            SemanticConvetion.GEN_AI_SYSTEM_VLLM,
+                        SemanticConvetion.GEN_AI_ENVIRONMENT:
+                            environment,
+                        SemanticConvetion.GEN_AI_TYPE:
+                            SemanticConvetion.GEN_AI_TYPE_CHAT,
+                        SemanticConvetion.GEN_AI_REQUEST_MODEL:
+                            model
+                    }
+
+                    metrics["genai_requests"].add(1, attributes)
+                    metrics["genai_total_tokens"].add(total_tokens, attributes)
+                    metrics["genai_completion_tokens"].add(output_tokens, attributes)
+                    metrics["genai_prompt_tokens"].add(input_tokens, attributes)
+                    metrics["genai_cost"].record(cost, attributes)
+
+                # Return original response
+                return response
+
+            except Exception as e:
+                handle_exception(span, e)
+                logger.error("Error in trace creation: %s", e)
+
+                # Return original response
+                return response
+
+    return wrapper

openlit/semcov/__init__.py

@@ -101,6 +101,7 @@ class SemanticConvetion:
     GEN_AI_SYSTEM_OLLAMA = "ollama"
     GEN_AI_SYSTEM_GPT4ALL = "gpt4all"
     GEN_AI_SYSTEM_ELEVENLABS = "elevenlabs"
+    GEN_AI_SYSTEM_VLLM = "vLLM"
     GEN_AI_SYSTEM_LANGCHAIN = "langchain"
     GEN_AI_SYSTEM_LLAMAINDEX = "llama_index"
     GEN_AI_SYSTEM_HAYSTACK = "haystack"

{openlit-1.17.0.dist-info → openlit-1.18.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: openlit
-Version: 1.17.0
+Version: 1.18.1
 Summary: OpenTelemetry-native Auto instrumentation library for monitoring LLM Applications, facilitating the integration of observability into your GenAI-driven projects
 Home-page: https://github.com/openlit/openlit/tree/main/openlit/python
 Keywords: OpenTelemetry,otel,otlp,llm,tracing,openai,anthropic,claude,cohere,llm monitoring,observability,monitoring,gpt,Generative AI,chatGPT
@@ -68,6 +68,8 @@ This project adheres to the [Semantic Conventions](https://github.com/open-telem
 | [✅ Vertex AI](https://docs.openlit.io/latest/integrations/vertexai)         |                                                                          |                                       |               |
 | [✅ Groq](https://docs.openlit.io/latest/integrations/groq)                  |                                                                          |                                       |               |
 | [✅ ElevenLabs](https://docs.openlit.io/latest/integrations/elevenlabs)      |                                                                          |                                       |               |
+| [✅ vLLM](https://docs.openlit.io/latest/integrations/vllm)                  |                                                                          |                                       |               |
+
 ## Supported Destinations
 - [✅ OpenTelemetry Collector](https://docs.openlit.io/latest/connections/otelcol)
 - [✅ Prometheus + Tempo](https://docs.openlit.io/latest/connections/prometheus-tempo)

{openlit-1.17.0.dist-info → openlit-1.18.1.dist-info}/RECORD

@@ -1,5 +1,5 @@
 openlit/__helpers.py,sha256=lrn4PBs9owDudiCY2NBoVbAi7AU_HtUpyOj0oqPBsPY,5545
-openlit/__init__.py,sha256=eJKH1Op7wzBsuoBYuM_C022Jo7cCtRQBJxf2lpDfe_o,14981
+openlit/__init__.py,sha256=LfU5w-D62u5pY70DdNbv5_DGtqeL1Yb0TlY-l0NAn8I,15103
 openlit/instrumentation/anthropic/__init__.py,sha256=oaU53BOPyfUKbEzYvLr1DPymDluurSnwo4Hernf2XdU,1955
 openlit/instrumentation/anthropic/anthropic.py,sha256=y7CEGhKOGHWt8G_5Phr4qPJTfPGRJIAr9Yk6nM3CcvM,16775
 openlit/instrumentation/anthropic/async_anthropic.py,sha256=Zz1KRKIG9wGn0quOoLvjORC-49IvHQpJ6GBdB-4PfCQ,16816
@@ -32,8 +32,8 @@ openlit/instrumentation/mistral/__init__.py,sha256=zJCIpFWRbsYrvooOJYuqwyuKeSOQL
 openlit/instrumentation/mistral/async_mistral.py,sha256=WXU46nbe61IglfGehrwdprMVB6hNiuqqmJHE3XvvP0E,22192
 openlit/instrumentation/mistral/mistral.py,sha256=DC-wLIypokPEEAbVSKX5sytv94DY2QDnk12401e5vq8,22039
 openlit/instrumentation/ollama/__init__.py,sha256=cOax8PiypDuo_FC4WvDCYBRo7lH5nV9xU92h7k-eZbg,3812
-openlit/instrumentation/ollama/async_ollama.py,sha256=UBwl-Jv4VPyPuhKIesL0VYIQ2u1AZ0U8c9MZQzEPn6c,30594
-openlit/instrumentation/ollama/ollama.py,sha256=bZPgkJJUD3RgJTUUt98cklfZ8Y1lsSgQGOUXXHk53BI,30504
+openlit/instrumentation/ollama/async_ollama.py,sha256=7lbikD-I9k8VL63idqj3VMEfiEKJmFNUPR8Xb6g2phQ,31366
+openlit/instrumentation/ollama/ollama.py,sha256=lBt1d3rFnF1tFbfdOccwjEafHnmTAUGsiOKSHku6Fkw,31277
 openlit/instrumentation/openai/__init__.py,sha256=AZ2cPr3TMKkgGdMl_yXMeSi7bWhtmMqOW1iHdzHHGHA,16265
 openlit/instrumentation/openai/async_azure_openai.py,sha256=XbST1UE_zXzNL6RX2XwCsK_a6IhG9PHVTMKBjGrUcB0,48961
 openlit/instrumentation/openai/async_openai.py,sha256=RGNpKLsHYfJXjj1ImuWRJToVSs0wdvMNp2kyTBrBaDw,47578
@@ -48,10 +48,12 @@ openlit/instrumentation/transformers/transformers.py,sha256=KNAT2ROjziW6OAP6Y0Ec
 openlit/instrumentation/vertexai/__init__.py,sha256=N3E9HtzefD-zC0fvmfGYiDmSqssoavp_i59wfuYLyMw,6079
 openlit/instrumentation/vertexai/async_vertexai.py,sha256=8JwSwLPPA4lAatf4w_5kJ5_YZDLwl5yG8N59cTD-EZM,55198
 openlit/instrumentation/vertexai/vertexai.py,sha256=R6dDQfC3YFoZDygxU2fkflcMsqIv8AVoU3XOwWSvpwA,54951
+openlit/instrumentation/vllm/__init__.py,sha256=OVWalQ1dXvip1DUsjUGaHX4J-2FrSp-T-qCVOfw7OZo,1495
+openlit/instrumentation/vllm/vllm.py,sha256=lDzM7F5pgxvh8nKL0dcKB4TD0Mc9wXOWeXOsOGN7Wd8,6527
 openlit/otel/metrics.py,sha256=O7NoaDz0bY19mqpE4-0PcKwEe-B-iJFRgOCaanAuZAc,4291
 openlit/otel/tracing.py,sha256=vL1ifMbARPBpqK--yXYsCM6y5dSu5LFIKqkhZXtYmUc,3712
-openlit/semcov/__init__.py,sha256=7E8qfoNRQbu6QAYBOkSUlESOQ8-VzvLTIKyj75BrHsY,7466
-openlit-1.17.0.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-openlit-1.17.0.dist-info/METADATA,sha256=mY_exhD2wObnFejJJJjXxlMwdMZ34uZBKlObiwhMc1Y,14120
-openlit-1.17.0.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-openlit-1.17.0.dist-info/RECORD,,
+openlit/semcov/__init__.py,sha256=EvoNOKtc7UKwLZ3Gp0-B1zwmeTcAIbx8O7wvAw8wXP4,7498
+openlit-1.18.1.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+openlit-1.18.1.dist-info/METADATA,sha256=l-nXpcuMqzXPGB-IiwqWhvkjq-pW4lsK_8Q5_rH3uoo,14334
+openlit-1.18.1.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+openlit-1.18.1.dist-info/RECORD,,