Arize Phoenix

Overview

Phoenix helps you understand and improve AI applications by giving you a workflow for debugging and iteration. You can send detailed logging information, known as traces, from your app to see exactly what happened during a run, score outputs using evaluation tests to identify failures and regressions, iterate on your prompts using real production examples, and optimize your app with experiments that compare changes on the same inputs. Phoenix is built by Arize AI and the open-source community, on top of OpenTelemetry and powered by OpenInference instrumentation.

Key Features

• Tracing: See what happened during a single run of your AI application, step by step, capturing model calls, retrieval, tool use, and custom logic.
• Evaluation: Score traces and spans with LLM-based evaluators, code-based checks, or human labels to track performance and identify failures consistently.
• Prompt Engineering: Version prompts, test prompt variants across datasets, and replay calls to see how changes affect outputs before rolling them out.
• Datasets & Experiments: Group traces into datasets, rerun them through different versions of your application, and compare evaluation results to confirm whether a change actually improved performance.
• PXI (Phoenix Intelligence): An AI engineering agent built into Phoenix for debugging traces, iterating on prompts, and navigating the product.
• Remote MCP Server: Connect Claude Code, Cursor, and other MCP clients directly to your Phoenix instance's /mcp endpoint to query traces, datasets, experiments, and more.

Use Cases

• Debugging LLM applications by inspecting traces to understand behavior and identify where time is spent.
• Measuring output quality with evaluations to track performance and identify failures consistently.
• Iterating on prompts using real production examples and comparing prompt variants across datasets.
• Optimizing applications with experiments that compare changes on the same inputs to confirm improvements.

Getting Started

• Start Phoenix in one terminal: `uvx arize-phoenix serve` (or `phoenix serve` after pip install). It serves at http://localhost:6006\.
• Add tracing with a coding agent: run `npx -y @arizeai/phoenix-cli setup` or `npm i -g @arizeai/phoenix-cli && px setup` from your app's root directory.
• Alternatively, follow the Python or TypeScript quick starts to send traces, measure performance, iterate on prompts, and optimize with experiments.

Deployment & Requirements

• Phoenix container images are available via Docker Hub and can be deployed using Docker or Kubernetes via the Helm chart.
• Deployment options include Docker Compose, Kubernetes/Helm, and cloud platforms like Railway, Render, Google Cloud, Azure, and AWS.
• The Google Cloud button builds Phoenix from source in Cloud Shell rather than deploying the prebuilt Docker Hub image.
• The Azure template serves plain HTTP (Azure Container Instances does not terminate TLS) — front it with a TLS proxy such as an Application Gateway before production use.

Before You Adopt

• License: Other. Review its terms before using, modifying, or distributing the project.
• The Azure deployment template serves plain HTTP and requires a TLS proxy for production use.