Arize Phoenix is an open-source platform for tracing, evaluating, and debugging LLM apps and AI agents, built on OpenTelemetry. Free to self-host.
Category
AI Infrastructure & MLOps
Pricing
Free (open source) + paid managed cloud tier, from Free
Verified
Not yet
Last updated
July 18, 2026
Founded
2020
Headquarters
Berkeley, California
Free PlanWeb AppAPIOpen SourceAISelf-Hosted
Overview
Arize Phoenix is an open-source AI observability and evaluation platform built by Arize AI, an AI observability company founded in 2020 with offices in Berkeley, California and New York City. Phoenix launched publicly in April 2023.
It helps engineers trace, evaluate, and debug LLM applications and AI agents by capturing distributed traces of every model call, retrieval step, and tool invocation.
Key Features
Phoenix is built on OpenTelemetry (OTLP) and Arize's OpenInference instrumentation standard, with auto-instrumentation for frameworks like LangChain, LlamaIndex, DSPy, CrewAI, and LangGraph, plus providers like OpenAI, Anthropic, Google, and AWS Bedrock.
Beyond tracing, it supports LLM-based, code-based, and human-label evaluations, and lets teams build experiment datasets from real production traces to iterate on prompts and catch regressions.
Pricing
Phoenix's open-source core is free under an ELv2 license and can be self-hosted locally, in Docker, or in Kubernetes. Arize also offers Arize AX, a paid managed cloud version that includes two free Phoenix Cloud instances to start, with paid tiers for production-scale deployments.
Key Features
Distributed tracing — Captures every LLM call, retrieval, and agent tool step for full visibility into execution.
Evaluations — Score traces and outputs with LLM-based evaluators, code-based checks, or human labels.
Experimentation — Build datasets from real traces and run experiments to test and compare changes.
Native OpenTelemetry support — Vendor- and language-agnostic tracing built on OTLP and OpenInference.
Broad framework auto-instrumentation — Out-of-the-box support for LangChain, LlamaIndex, DSPy, CrewAI, LangGraph, and more.
Flexible deployment — Run locally, in a notebook, via Docker/Kubernetes, or through managed Arize AX cloud.
Pros & Cons
Pros
Fully open source and self-hostable
Framework- and vendor-agnostic via OpenTelemetry
Active open-source community with 9,000+ GitHub stars
Free managed cloud tier available to start (Arize AX)
Cons
Pricing for production-scale Arize AX deployments isn't publicly listed
Self-hosting and scaling requires infrastructure setup and familiarity with OpenTelemetry
Yes, the open-source core is free under the ELv2 license and can be self-hosted at no cost.
What's the difference between Phoenix and Arize AX?
Phoenix OSS is self-hosted and free; Arize AX is Arize's managed cloud version, offering free starter instances and paid tiers for production-scale use.
What frameworks does Phoenix integrate with?
LangChain, LlamaIndex, DSPy, CrewAI, LangGraph, the Vercel AI SDK, the OpenAI Agents SDK, the Claude Agent SDK, and more.
Who makes Arize Phoenix?
Arize AI, an AI observability company founded in 2020, headquartered in Berkeley, California.