# OpenLIT

> Open-source, OpenTelemetry-native platform for AI agents: tracing, evals, guardrails, prompt hub, vault, and cost and GPU monitoring; self-host free.

- id: `openlit`
- category: Evals & Observability (`evals-observability`)
- url: https://openlit.io
- repo: https://github.com/openlit/openlit
- docs: https://docs.openlit.io
- status: active
- pricing: open-source
- license: Apache-2.0
- tags: observability, tracing, opentelemetry, evals, guardrails, self-hosted, gpu-monitoring
- llms.txt: https://docs.openlit.io/llms.txt
- related: OpenLLMetry (https://indexagentica.com/entries/openllmetry/); OpenTelemetry GenAI Semantic Conventions (https://indexagentica.com/entries/otel-genai-semconv/); Langfuse (https://indexagentica.com/entries/langfuse/); Arize Phoenix (https://indexagentica.com/entries/arize-phoenix/); Opik (https://indexagentica.com/entries/opik/)
- sources: https://github.com/openlit/openlit , https://docs.openlit.io/llms.txt
- added: 2026-10-22
- updated: 2026-10-22
- last verified: 2026-10-22
- maintainer: OpenLIT
- submitted by: agentica-curator
- page: https://indexagentica.com/entries/openlit/
- json: https://indexagentica.com/api/entries/openlit.json

Traces every LLM call, tool call, MCP request and retrieval via OpenTelemetry SDKs (`pip install openlit` / `npm install openlit`), evaluates on real traces (LLM-as-a-judge, programmatic, human feedback, CI gates), adds runtime guardrails (prompt injection, sensitive or restricted topics) and manages prompts and secrets without redeploying. Tracks cost (incl. custom pricing) and NVIDIA/AMD/Intel GPUs. Self-host with `docker compose up -d`.