极光AI可观测性平台

Phoenix is an open-source AI observability platform designed for experimentation, evaluation, and troubleshooting. It provides:

• Tracing - Trace your LLM application's runtime using OpenTelemetry-based instrumentation.
• Evaluation - Leverage LLMs to benchmark your application's performance using response and retrieval evals.
• Datasets - Create versioned datasets of examples for experimentation, evaluation, and fine-tuning.
• Experiments - Track and evaluate changes to prompts, LLMs, and retrieval.
• Playground- Optimize prompts, compare models, adjust parameters, and replay traced LLM calls.
• Prompt Management- Manage and test prompt changes systematically using version control, tagging, and experimentation.

Phoenix is vendor and language agnostic with out-of-the-box support for popular frameworks (🦙LlamaIndex, 🦜⛓LangChain, Haystack, 🧩DSPy, 🤗smolagents) and LLM providers (OpenAI, Bedrock, MistralAI, VertexAI, LiteLLM, Google GenAI and more). For details on auto-instrumentation, check out the OpenInference project.

Phoenix runs practically anywhere, including your local machine, a Jupyter notebook, a containerized deployment, or in the cloud.

Installation

Install Phoenix via pip or conda

pip install arize-phoenix

Phoenix container images are available via Docker Hub and can be deployed using Docker or Kubernetes. Arize AI also provides cloud instances at app.phoenix.arize.com.

Packages

The arize-phoenix package includes the entire Phoenix platfom. However if you have deployed the Phoenix platform, there are light-weight Python sub-packages and TypeScript packages that can be used in conjunction with the platfrom.

Python Subpackages

| Package | Version & Docs | Description | | --------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------ | | arize-phoenix-otel | | Provides a lightweight wrapper around OpenTelemetry primitives with Phoenix-aware defaults | | arize-phoenix-client | | Lightweight client for interacting with the Phoenix server via its OpenAPI REST interface | | arize-phoenix-evals | | Tooling to evaluate LLM applications including RAG relevance, answer relevance, and more |

TypeScript Subpackages

| Package | Version & Docs | Description | | --------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------ | | @arizeai/phoenix-otel | | Provides a lightweight wrapper around OpenTelemetry primitives with Phoenix-aware defaults | | @arizeai/phoenix-client | | Client for the Arize Phoenix API | | @arizeai/phoenix-evals | | TypeScript evaluation library for LLM applications (alpha release) | | @arizeai/phoenix-mcp | | MCP server implementation for Arize Phoenix providing unified interface to Phoenix's capabilities | | @arizeai/phoenix-cli | | CLI for fetching traces, datasets, and experiments for use with Claude Code, Cursor, and other coding agents |

Tracing Integrations

Phoenix is built on top of OpenTelemetry and is vendor, language, and framework agnostic. For details about tracing integrations and example applications, see the OpenInference project.

Python Integrations | Integration | Package | Version Badge | |------------------|-----------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------| | OpenAI | openinference-instrumentation-openai | | | OpenAI Agents | openinference-instrumentation-openai-agents | | | LlamaIndex | openinference-instrumentation-llama-index | | | DSPy | openinference-instrumentation-dspy | | | AWS Bedrock | openinference-instrumentation-bedrock | | | LangChain | openinference-instrumentation-langchain | | | MistralAI | openinference-instrumentation-mistralai | | | Google GenAI | openinference-instrumentation-google-genai | | | Google ADK | openinference-instrumentation-google-adk | | | Guardrails | openinference-instrumentation-guardrails | | | VertexAI | openinference-instrumentation-vertexai | | | CrewAI | openinference-instrumentation-crewai | | | Haystack | openinference-instrumentation-haystack | | | LiteLLM | openinference-instrumentation-litellm | | | Groq | openinference-instrumentation-groq | | | Instructor | openinference-instrumentation-instructor | | | Anthropic | openinference-instrumentation-anthropic | | | Smolagents | openinference-instrumentation-smolagents | | | Agno | openinference-instrumentation-agno | | | MCP | openinference-instrumentation-mcp | | | Pydantic AI | openinference-instrumentation-pydantic-ai | | | Autogen AgentChat | openinference-instrumentation-autogen-agentchat | | | Portkey | openinference-instrumentation-portkey | |

Span Processors

Normalize and convert data across other instrumentation libraries by adding span processors that unify data.

| Package | Description | Version | | ----------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | openinference-instrumentation-openlit | OpenInference Span Processor for OpenLIT traces. | | | openinference-instrumentation-openllmetry | OpenInference Span Processor for OpenLLMetry (Traceloop) traces. | |

JavaScript Integrations

| Integration | Package | Version Badge | | ------------------------------------------------------------------------------------------ | -------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | OpenAI | @arizeai/openinference-instrumentation-openai | | | LangChain.js | @arizeai/openinference-instrumentation-langchain | | | Vercel AI SDK | @arizeai/openinference-vercel | | | BeeAI | @arizeai/openinference-instrumentation-beeai | | | Mastra | @mastra/arize | |

Java Integrations

| Integration | Package | Version Badge | | --------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | LangChain4j | openinference-instrumentation-langchain4j | | | SpringAI | openinference-instrumentation-springAI | |

Platforms

| Platform | Description | Docs | | -------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- | | BeeAI | AI agent framework with built-in observability | Integration Guide | | Dify | Open-source LLM app development platform | Integration Guide | | Envoy AI Gateway | AI Gateway built on Envoy Proxy for AI workloads | Integration Guide | | LangFlow | Visual framework for building multi-agent and RAG applications | Integration Guide | | LiteLLM Proxy | Proxy server for LLMs | Integration Guide |

Security & Privacy

We take data security and privacy very seriously. For more details, see our Security and Privacy documentation.

Telemetry

By default, Phoenix collects basic web analytics (e.g., page views, UI interactions) to help us understand how Phoenix is used and improve the product. None of your trace data, evaluation results, or any sensitive information is ever collected.

You can opt-out of telemetry by setting the environment variable: PHOENIX_TELEMETRY_ENABLED=false

Community

Join our community to connect with thousands of AI builders.

• 🌍 Join our Slack community.
• 📚 Read our documentation.
• 💡 Ask questions and provide feedback in the #phoenix-support channel.
• 🌟 Leave a star on our GitHub.
• 🐞 Report bugs with GitHub Issues.
• 𝕏 Follow us on 𝕏.
• 🗺️ Check out our roadmap to see where we're heading next.
• 🧑‍🏫 Deep dive into everything Agents and LLM Evaluations on Arize's Learning Hubs.

Breaking Changes

See the migration guide for a list of breaking changes.

Copyright, Patent, and License

Copyright 2025 Arize AI, Inc. All Rights Reserved.

Portions of this code are patent protected by one or more U.S. Patents. See the IP_NOTICE.

This software is licensed under the terms of the Elastic License 2.0 (ELv2). See LICENSE.