OpenLLMetry
Open-source observability for LLM apps, built on OpenTelemetry.
About
OpenLLMetry is an open-source observability framework for LLM and generative AI applications built on OpenTelemetry, maintained by Traceloop under Apache 2.0. It provides drop-in instrumentation for 30+ LLM providers, vector databases, and agent frameworks, emitting standardized GenAI traces over OTLP to any observability backend — Datadog, Grafana, Honeycomb, Langfuse, and others.
Its semantic conventions for LLM telemetry have been upstreamed into the OpenTelemetry GenAI semantic conventions, making it the de facto standard vocabulary for GenAI spans. The simplest setup is a single Traceloop.init() call via the traceloop-sdk, though wiring traces into an existing OTel collector takes a bit more configuration.
What it can do
- Drop-in instrumentation for 30+ LLM providers
- Vector DB and agent framework instrumentations
- Standard OTLP export to any backend
- GenAI semantic conventions upstreamed to OTel
- One-line SDK initialization
Setup
pip install traceloop-sdk, then call Traceloop.init() at the start of your application.
Prerequisites
An OpenTelemetry-compatible backend or collector to receive traces.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 79; Trending score 82/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.