May 08, 2026
Arize Phoenix is the open-source community's answer to the need for deep visibility into LLM applications. It provides a notebook-first experience for tracing, evaluating, and visualizing the performance of your RAG pipelines and AI agents.
Phoenix captures full traces of your LLM calls, including retrieval steps and tool usage. It allows you to visualize these traces in a web UI, making it easy to spot where a RAG system retrieved the wrong document or where an LLM hallucinated based on the provided context.
One of Phoenix's unique features is its ability to visualize and analyze your high-dimensional embeddings. By mapping your data into a 3D space, you can identify "clusters" of failures or areas where your vector search is underperforming, allowing for targeted improvements to your AI knowledge base.