arize-phoenix-client

What it is and what it does

Phoenix Client is a Python SDK for interacting with the Arize Phoenix observability platform via REST API. It abstracts away HTTP details and provides resource-oriented methods for managing LLM application data: creating and versioning prompts with template variables, building evaluation datasets from DataFrames or dictionaries, querying and filtering traces, recording experiment results, and collecting human feedback and automated evaluations.

The package is designed for teams running LLM applications who need to track behavior, debug issues, and run structured evaluations programmatically. It supports both synchronous and asynchronous clients, discovers configuration from environment variables or a `.env.phoenix` dotenv file, and integrates with pytest for running LLM evaluations as ordinary test cases. Its runtime dependencies (httpx, OpenTelemetry SDK and exporters, openinference conventions, tqdm, typing-extensions) are all lightweight and widely used.

Use it for:

• Create and version prompt templates with variable substitution, then retrieve and format them for use with LLM APIs.
• Build evaluation datasets from pandas DataFrames or CSV files and manage them centrally for reproducible experiment runs.
• Query and filter application traces by time range, status, or custom criteria to debug LLM behavior in production.
• Run LLM evaluations as pytest tests and automatically record results as Phoenix experiments for tracking.
• Collect human feedback and automated evaluation scores on LLM outputs and associate them with specific traces.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Phoenix Client provides a Python interface to the Phoenix observability platform's REST API, enabling programmatic management of datasets, experiments, traces, prompts, and feedback for LLM applications.

Yes. The package is actively maintained, has no known vulnerabilities, uses a permissive license, and offers low install friction. Install it if you are already running a Phoenix instance and need programmatic access to its API from Python; it is purpose-built for that workflow. If you do not have a Phoenix server running or do not need API-level control over datasets and experiments, it will not be useful.

Install

arize-phoenix-client on PyPI

pip install arize-phoenix-client

uv add arize-phoenix-client

poetry add arize-phoenix-client

Installing arize-phoenix-client

Before you install

Low install friction with a pure-Python wheel distribution. Active maintenance with a recent release (3 days old) and steady commit activity. Supports modern Python versions (3.10–3.14).

License in practice

Apache-2.0 permissive license allows commercial and private use with minimal restrictions; you must retain license notices and provide a copy of the license.

Quickstart

pip install arize-phoenix-client

from phoenix.client import Client

client = Client(base_url="http://localhost:6006")
traces = client.traces.get_traces(project_identifier="my-llm-app")

Requires a running Phoenix server instance (local or remote) at the configured endpoint; PHOENIX_ENDPOINT environment variable or explicit base_url parameter must be set.

Verify before relying

• Whether the pytest integration (Eval CI) requires additional setup beyond the optional [pytest] extra.
• Performance characteristics when working with large trace datasets or high-volume queries.
• Exact retry and timeout behavior for network failures against the Phoenix API.

Package facts

Evidence: arize_phoenix_client-3.1.0-py3-none-any.whl

Keywords: Explainability, Monitoring, Observability

Tags

More Monitoring packages