OpenLLMetry is an open-source extension of OpenTelemetry for observing large language model applications. It emits standard telemetry while adding LLM-specific visibility across provider calls, APIs, databases, and vector databases. section 11.4.7
OpenLLMetry supports production observability by enabling teams to:
- Trace LLM executions and inspect interactions with providers such as OpenAI or Anthropic. section 11.4.7
- Monitor generic operations, including API, database, and vector-database interactions. section 11.4.7
- Relate completion-token counts to request latency and inspect prompt or response details, subject to privacy and legal controls. section 11.4.7
Instrumenting an Azure OpenAI test
Install the Traceloop SDK, obtain an API key, and call Traceloop.init(api_key=TRACELOOP_API_KEY) before creating an AzureOpenAI client. A load-test script can then send repeated conversational inputs for observability.
