Arize Phoenix

AI Observability and Evaluation

Phoenix is an open-source AI development platform for experimentation, evaluation, and troubleshooting. Trace your application, evaluate its outputs, iterate on prompts, and ship with confidence.

Phoenix is built by Arize AI and the open-source community. It is built on top of OpenTelemetry and is powered by OpenInference instrumentation. See Integrations for details.

Features

Tracing

Tracing in Phoenix

Tracing gives you visibility into every step of your LLM application—model calls, retrieval, tool use, and more.

Phoenix accepts traces over OpenTelemetry (OTLP) and provides auto-instrumentation for popular frameworks (LlamaIndex, LangChain, DSPy, Mastra, Vercel AI SDK), providers (OpenAI, Bedrock, Anthropic), and languages (Python, TypeScript, Java).

Evaluation

Evals in the Phoenix UI

Evaluate your application to understand its true performance:

• LLM-based evaluations — Run pre-built or custom evaluators on your data
• Evaluator integrations — Use Phoenix evals, or bring your own from Ragas, Deepeval, or Cleanlab
• Human annotations — Attach ground truth labels directly in the UI

Prompt Engineering

Phoenix Prompt Playground

Streamline your prompt engineering workflow:

• Prompt Management — Version, store, and deploy prompts
• Prompt Playground — Experiment with prompts and models side-by-side
• Span Replay — Debug by replaying LLM calls with different inputs
• Prompts in Code — Sync prompts across environments via SDK

Datasets & Experiments

Experiments in Phoenix

Datasets & Experiments help you test systematically and track improvements:

• Run Experiments — Compare different versions of your application
• Create Datasets — Collect traces or upload from code/CSV
• Test at Scale — Run datasets through Playground or export for fine-tuning

Quick Starts

Running Phoenix for the first time? Select a quick start below.

Tracing

See what’s happening inside your LLM application with distributed tracing

Prompt Engineering

Experiment with prompts, compare models, and version your work

Datasets & Experiments

Test your application systematically and track performance over time

Evaluation

Measure quality with LLM-as-a-judge and custom evaluators

Next Steps

Integrations

Add instrumentation for OpenAI, LangChain, LlamaIndex, and more

Self-Host

Deploy Phoenix on Docker, Kubernetes, or your cloud of choice

Tutorials

Example notebooks for tracing, evals, RAG analysis, and more

Community

Join the Phoenix Slack to ask questions and connect with developers