The OpenLIT Operator supports multiple instrumentation providers, each offering different capabilities and approaches to AI observability. This overview helps you choose the right provider for your use case.

​
What is Instrumentation?

Instrumentation is the process of adding observability code to your applications to collect telemetry data such as:
  • Traces: Detailed execution flows showing request paths
  • Metrics: Quantitative measurements of performance and usage
  • Logs: Structured event records for debugging and monitoring
  • Costs: Token usage and estimated API costs
The OpenLIT Operator automatically injects instrumentation code into your applications without requiring code changes.

​
Supported Providers

The operator supports four instrumentation providers:

OpenLIT

Complete AI observability built on OpenTelemetry for LLMs, vector databases, GPUs and AI Agents with distributed tracing and metrics

OpenInference

OpenInference is a set of conventions and plugins that is complimentary to OpenTelemetry to enable tracing of AI applications.

OpenLLMetry

OpenLLMetry is a set of extensions built on top of OpenTelemetry that gives you complete observability over your LLM application.

Custom

Bring your own instrumentation image with custom implementations

​
How Instrumentation Works

1

Webhook Interception

The operator intercepts pod creation requests via an admission webhook
2

Label Matching

Checks if the pod matches any AutoInstrumentation selector labels
3

Init Container Injection

Injects an init container with the selected instrumentation provider
4

Environment Setup

The init container installs instrumentation packages and sets up environment variables
5

Application Start

Your application starts with instrumentation automatically enabled
6

Telemetry Collection

Traces, metrics, and costs are automatically collected and sent to your OTLP endpoint

​
Provider Comparison

FeatureOpenLITOpenInferenceOpenLLMetryCustom
AI Instrumentations50+30+30+Depends
Cost Tracking✅ Advanced✅ Basic✅ BasicDepends
Performance Metrics✅ Comprehensive✅ Standard✅ StandardDepends
Vendor Lock-inNoneNoneNoneNone

​
Troubleshooting

​
Common Issues