Skip to content
← Tools / Skills
ToolArize AI/10 trustMetrics-checked
Arize Phoenix logo

Arize Phoenix

Open-source LLM tracing and evaluation, OpenTelemetry-native, runs in a notebook.

Acquired — still operating

acquired by Dynatrace Aug 2026

Visit official site →

About

Arize Phoenix is an open-source AI observability and evaluation platform built on OpenTelemetry: trace every agent step, run LLM-as-judge evals with built-in evaluators (faithfulness, relevance, hallucination, toxicity), build datasets from traces, and run experiments — all self-hosted with your traces staying in your environment. It works out of the box with LangChain, LangGraph, LlamaIndex, CrewAI, DSPy, and the OpenAI Agents SDK.

It runs locally in a notebook with zero infrastructure (pip install arize-phoenix) or as a deployed server, and Arize offers hosted Phoenix Cloud instances. ~11k GitHub stars, ELv2 licensed.

Honest limits: Arize was acquired by Dynatrace in August 2026, so the product's direction may shift; and its deepest strengths are RAG/agent debugging rather than full production APM.

What it can do

  • OTel-native tracing for agents and RAG
  • Built-in evaluators (faithfulness, relevance, toxicity)
  • Datasets and experiment tracking
  • Embedding analysis and drift views
  • Prompt playground
  • Notebook-local or self-hosted

Setup

  1. Install: pip install arize-phoenix.
  2. Launch: phoenix.trace_notebook() or run the server.
  3. Instrument with @trace decorators or auto-instrumentation for your framework.
  4. Inspect traces in the UI, then build evals and datasets from them.

Prerequisites

Python 3.9+; an LLM app to instrument.

Use cases

dataautomation

Works with

langchainlanggraphllamaindexcrewaidspyopenai-agents

Trust score — 8/10

/10 trustMetrics-checked

Scored 8/10 (metrics tier, metrics-v1). Evidence: ~11k GitHub stars; OpenTelemetry-native tracing; Arize acquired by Dynatrace Aug 2026 — direction may shift; Runs locally in a notebook with zero infrastructure. License: ELv2 licensed, open source.

★ 11,000 GitHub stars

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
8.0/10
Maintenance · recency, releases, issue load
9.0/10
Openness · license + docs
8.0/10
Manual review · sanity read + red-flag check
7.0/10
  • · ~11k GitHub stars; OpenTelemetry-native tracing
  • · Arize acquired by Dynatrace Aug 2026 — direction may shift
  • · Runs locally in a notebook with zero infrastructure

Sources

Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.