Phoenix
Open-source AI observability and evaluation platform for tracing, debugging, and improving agents and LLM applications.
About
Phoenix by Arize is an open-source platform built for AI engineers who need visibility into how their agents and LLM pipelines behave. It captures detailed traces of every step an agent takes, so you can pinpoint exactly where a workflow went wrong. Built-in evaluation tools let you score outputs, run experiments on datasets drawn from real traces, and measure quality improvements before shipping changes. Phoenix is built on OpenTelemetry and works with any model or framework, and it can run locally, in Docker, or on Kubernetes. A hosted cloud option includes a free tier; larger deployments have paid plans.