Arize Phoenix

Arize Phoenix is an AI observability platform for experimenting with, evaluating and troubleshooting LLM applications and agents. It's self-hosted and runs on your machine, in a container or in the cloud; Arize also offers a managed platform, Arize AX.

Phoenix traces your app with OpenTelemetry-based instrumentation and uses LLMs to evaluate responses and retrieval. You can build versioned datasets, run experiments on changes to prompts, models and retrieval, and compare models in a playground that replays traced calls. It includes prompt management, a built-in AI agent for debugging traces, and a remote MCP server for Claude Code, Cursor and other clients. It works with many frameworks, including the OpenAI Agents SDK and Claude Agent SDK.

Phoenix gives teams the tracing and evaluation tools they need to improve LLM apps without sending data to a hosted service. It's under the Elastic License 2.0.

Arize Phoenix uses the Elastic License 2.0, a source-available license. You can use, modify and self-host it, including commercially, but you may not offer it to others as a hosted or managed service.

Stars

0

Forks

0

Commits

0

Repository

Arize-ai/phoenix

• LangSmithLangChain's hosted platform for tracing and evaluating LLM apps.
• WWeights & BiasesCommercial platform for ML and LLM experiment tracking.
• LangfuseNo. 211Open-source LLM engineering platform.

See every open-source alternative to LangSmith and Weights & Biases in the collection.