Arize Phoenix
Open-source LLM tracing and evaluation built on OpenTelemetry
What it is
An AI observability tool for experimenting with, evaluating and troubleshooting LLM applications. It collects traces through OpenTelemetry-based instrumentation and runs LLM-based evaluations, served from a local process started with pip or uvx. For engineers debugging agent and RAG behaviour during development.
Good for
- Inspect traces from an agent during development
- Score responses with LLM-based evaluators
- Debug retrieval quality in a RAG pipeline
Install or run
pip install arize-phoenixCommand as given in the project README. Read the repository before you run anything on your machine.
What you need
Python; an LLM API key for LLM-based evaluations
Sources
- https://github.com/Arize-ai/phoenix (project README, read 10 Nov 2026)
- https://api.github.com/repos/Arize-ai/phoenix (GitHub API, read 13 Nov 2026)
The summary is ours, written from the README. Spotted an error? See how we work.