Description
OpenLLMetry is an open-source observability library for large language model applications, built on OpenTelemetry. It records what your LLM calls are doing, so you can monitor performance, track cost and debug an AI application. The site says you can start with two lines of code, by initialising the SDK and marking the workflow functions you want traced. It is available for Python and TypeScript. The library supports many model providers, including OpenAI, Anthropic, Amazon Bedrock, Cohere, Gemini, Mistral AI, Ollama, together.ai and Watsonx, and frameworks such as LangChain, LlamaIndex and Chroma. Because it uses OpenTelemetry, the traces can be sent to the observability tools you already use, and the site lists destinations such as Datadog, Grafana Tempo, Honeycomb, New Relic, Dynatrace, Splunk and Instana. Traceloop, the company behind it, also offers a hosted platform, and the project's code is on GitHub.…
Preview
Highlights
- Two lines of code to start tracing
- Built on OpenTelemetry
- Traces for many LLM providers
- Works with LangChain and LlamaIndex
- Sends traces to your existing tools
Ratings & Reviews
No ratings yet
Be the first to rate OpenLLMetry.
People also visit
OpenLLMetry FAQ
Is OpenLLMetry free?
Yes. It is an open-source library, and its code is on GitHub.
How much code do I need to add?
The site says two lines: initialise the SDK and add a decorator to the workflow function you want to trace. Examples are given for Python and TypeScript.
Which models does it track?
The site lists OpenAI, Anthropic, Bedrock, Cohere, Gemini, Mistral AI, Ollama, together.ai and Watsonx.
Where do the traces go?
Because it is built on OpenTelemetry, you can send them to tools such as Datadog, Grafana Tempo, Honeycomb and New Relic, according to the site.
What are the best OpenLLMetry alternatives?
Popular alternatives to OpenLLMetry include OpenLIT: Zero code Agent Observability, Langtrace and Lmnr. See all OpenLLMetry alternatives


