Product / UX / Telemetry & observability

OpenLLMetry

OpenTelemetry-based instrumentation library for LLM applications. Auto-instruments LangChain, LlamaIndex, OpenAI SDK, and other popular frameworks. Exports telemetry data to any OTel-compatible backend (Jaeger, Grafana, Datadog, etc). The 'OpenTelemetry for LLMs' approach, standards-based, vendor-neutral. 7.1K GitHub stars.

traceloop-sdk v0.61.0 (released 31 May 2026, PyPI/GitHub); confirmed live June 2026. OTel-based instrumentation SDK; Traceloop hosted backend is the optional commercial destination.

Openness

5 high confidence

5.0

license: Apache-2.0(OSI)
source: public
core-gated: ungated

Apache-2.0 instrumentation library; exports to any OTel backend (Datadog, Honeycomb, etc.), so the OSS is genuinely standalone.

https://pypi.org/project/traceloop-sdk/ recorded 2026-06-04
License: Apache-2.0; v0.61.0 (31 May 2026)
https://github.com/traceloop/openllmetry recorded 2026-06-04
Apache-2.0; OTel-based GenAI instrumentation

Adoption

4 high confidence

4.0

traceloop-sdk ~3.15M PyPI downloads/month; 15+ LLM providers, 7 vector DBs, 25+ tested export destinations; ~7.2k stars corroborates only.

https://pypistats.org/packages/traceloop-sdk recorded 2026-06-04
~3,149,787 downloads last month
https://github.com/traceloop/openllmetry recorded 2026-06-04
~7.2k stars; 15+ providers, 25+ destinations

Capability

3 high confidence

3.0

Strong on tracing + integrations but narrow vs full platforms; it is an instrumentation layer, not an eval/prompt/dataset suite. Below the MLflow/Langfuse/Phoenix/Opik feature breadth (C4).

https://github.com/traceloop/openllmetry recorded 2026-06-04
instrumentation/integration-focused feature set; evals/prompt mgmt are Traceloop-platform features not OSS SDK

Unchanged since 2026-07-30 (last edited, not re-checked)