SDK configuration

All parameters for openlit.init() with their types, defaults, and environment variable equivalents.

Configure the OpenLIT SDK by passing parameters to openlit.init(), by setting environment variables before your process starts, or by using CLI arguments with openlit-instrument. Environment variables take precedence over CLI arguments, which take precedence over SDK parameters.

Environment variables take precedence over CLI arguments, which take precedence over SDK parameters passed directly to openlit.init().

Python

Manual instrumentation

import openlit

openlit.init(
    service_name="my-ai-app",
    environment="production",
    otlp_endpoint="http://127.0.0.1:4318",
)

Zero-code instrumentation (CLI)

export OTEL_SERVICE_NAME=my-ai-app
export OTEL_DEPLOYMENT_ENVIRONMENT=production
export OTEL_EXPORTER_OTLP_ENDPOINT=http://127.0.0.1:4318
openlit-instrument python your_app.py

Core parameters

service_name · string · default: "default"

The name of your service as it appears in traces and metrics. Maps to OTEL_SERVICE_NAME.

Replaces the deprecated application_name parameter.

environment · string · default: "default"

The deployment environment (for example, production, staging, development). Maps to OTEL_DEPLOYMENT_ENVIRONMENT.

otlp_endpoint · string · default: "None"

The OpenTelemetry OTLP endpoint to export traces and metrics to (for example, http://127.0.0.1:4318). Maps to OTEL_EXPORTER_OTLP_ENDPOINT.

When not set, OpenLIT initialises successfully but does not export any telemetry.

otlp_headers · dict · default: "None"

Authentication headers for your OTLP endpoint, for example {"Authorization": "Bearer <token>"}. Maps to OTEL_EXPORTER_OTLP_HEADERS.

tracer · opentelemetry.trace.Tracer · default: "None"

An existing OpenTelemetry Tracer instance. Pass this if you have already configured a tracer in your application and want OpenLIT to use the same provider. No environment variable equivalent.

event_logger · opentelemetry.logs.Logger · default: "None"

An existing OpenTelemetry EventLoggerProvider instance for emitting events. No environment variable equivalent.

meter · opentelemetry.metrics.Meter · default: "None"

An existing OpenTelemetry Meter instance. No environment variable equivalent.

disable_batch · boolean · default: "False"

Disables the OpenTelemetry batch span processor and sends spans immediately. Useful in development to see traces without delay. Maps to OPENLIT_DISABLE_BATCH.

Content and tracing parameters

capture_message_content · boolean · default: "True"

Captures LLM prompt and response content as span attributes. Disable this for privacy-sensitive workloads or to reduce span size. Maps to OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT.

max_content_length · int · default: "None"

Maximum character length for captured prompt and completion content. Content exceeding the limit is truncated with ... appended.

None (default), 0, or -1 disables truncation entirely. Maps to OPENLIT_MAX_CONTENT_LENGTH.

# Truncate all captured content to 500 characters
openlit.init(max_content_length=500)

detailed_tracing · boolean · default: "True"

Enables detailed component-level tracing for AI frameworks (for example, individual LangChain chain steps). When False, only workflow-level spans are created, reducing storage overhead. Maps to OPENLIT_DETAILED_TRACING.

disabled_instrumentors · list[str] · default: "None"

A list of instrumentor names to skip. Use this to disable instrumentation for specific libraries (for example, ["anthropic", "langchain"]). Maps to OPENLIT_DISABLED_INSTRUMENTORS (comma-separated).

Metrics parameters

disable_metrics · boolean · default: "False"

Disables all OpenTelemetry metrics collection, including cost tracking and performance metrics. Maps to OPENLIT_DISABLE_METRICS.

pricing_json · string · default: "None"

A file path or URL pointing to a custom pricing JSON file. Use this for fine-tuned or private models not in the built-in pricing database. Maps to OPENLIT_PRICING_JSON.

openlit.init(pricing_json="/path/to/custom-pricing.json")
# or
openlit.init(pricing_json="https://example.com/pricing.json")

System metrics

collect_system_metrics · boolean · default: "False"

Enables comprehensive system monitoring including CPU, memory, disk, network, and GPU metrics. Requires opentelemetry-instrumentation-system-metrics. GPU monitoring is automatically enabled if a supported GPU is detected. Maps to OPENLIT_COLLECT_SYSTEM_METRICS.

Evaluation export

evals_logs_export · boolean · default: "True"

When True, evaluation results are emitted as OTel Log Records. When False, they are emitted as OTel Events. Use True (the default) for backends like Grafana Loki that handle logs better than events. Maps to OPENLIT_EVALS_LOGS_EXPORT.

Database instrumentation parameters

These parameters apply to database instrumentations such as PostgreSQL (psycopg3).

capture_parameters · boolean · default: "False"

Captures database query parameters (the values bound to $1, $2, and so on) in spans.

Enabling this may expose sensitive data such as passwords, API keys, or personal information in your traces. Only enable in development environments or when you are certain parameters contain no sensitive data.

Maps to OPENLIT_CAPTURE_PARAMETERS.

enable_sqlcommenter · boolean · default: "False"

Injects OpenTelemetry trace context as SQL comments in queries, enabling correlation between application traces and database logs.

-- With SQLCommenter enabled
SELECT * FROM users WHERE id = $1
/*traceparent='00-abc123-789xyz-01',application='my_app'*/;

Maps to OPENLIT_ENABLE_SQLCOMMENTER.

Deprecated parameters

Parameter	Replacement	Notes
application_name	service_name	Still accepted for backward compatibility
collect_gpu_stats	collect_system_metrics	Use collect_system_metrics=True instead

TypeScript

Manual instrumentation

import Openlit from "openlit";

Openlit.init({
  applicationName: "my-ai-app",
  environment: "production",
  otlpEndpoint: "http://127.0.0.1:4318",
});

Configuration parameters

Parameter	Type	Default	Description
applicationName	string	"default"	Service name for tracing. Maps to OTEL_SERVICE_NAME.
environment	string	"default"	Deployment environment. Maps to OTEL_DEPLOYMENT_ENVIRONMENT.
otlpEndpoint	string	undefined	OTLP endpoint URL. Maps to OTEL_EXPORTER_OTLP_ENDPOINT.
otlpHeaders	object	undefined	Authentication headers. Maps to OTEL_EXPORTER_OTLP_HEADERS.
tracer	Tracer	undefined	Existing OTel Tracer instance.
disableBatch	boolean	true	Disable batch span processing.
traceContent	boolean	true	Capture LLM prompt and response content.
disabledInstrumentations	string[]	undefined	List of instrumentations to skip.
instrumentations	object	undefined	Manual instrumentation modules for patching.
pricing_json	string	built-in URL	Custom pricing JSON URL for cost tracking.

Resource attributes

Use standard OpenTelemetry environment variables to attach additional metadata to all telemetry:

export OTEL_RESOURCE_ATTRIBUTES="service.version=2.1.0,team=ai-platform"
export OTEL_SERVICE_VERSION=2.1.0

Environment variable	Description	Example
OTEL_RESOURCE_ATTRIBUTES	Key-value pairs for resource attributes	service.version=1.0.0
OTEL_SERVICE_VERSION	Service version	1.2.3
OTEL_RESOURCE_ATTRIBUTES_POD_NAME	Kubernetes pod name	my-app-pod-xyz
OTEL_RESOURCE_ATTRIBUTES_NODE_NAME	Kubernetes node name	node-123

Quickstart

Get your first trace running in minutes

Tracing

Custom spans, content capture, and manual tracing options

Integrations

60+ supported integrations