Arize Phoenix
Evaluation & observability · Open source
About
An open-source tool for tracing, evaluating and experimenting with LLM and agent applications, built on OpenTelemetry.
Key features
- OpenTelemetry tracing
- LLM-as-judge evals
- Prompt playground
- Datasets and experiments
Best for
- — Local debugging of LLM traces
- — OpenTelemetry-based stacks
- Pricing
- Open source
- Platforms
- Self-hosted, Web, API
- Works with
- LangChain, LlamaIndex, OpenAI
- Last checked
- 2026-10-29
- Status
- Listed, not yet tested
Our take
Standards-based and free to run locally. Enterprise features live in Arize's paid platform.
Features and pricing change often — confirm on arize.com before you commit.
Alternatives to Arize Phoenix
| Tool | Type | Pricing | Best for |
|---|---|---|---|
| Arize Phoenix (this) | Evaluation & observability | Open source | Local debugging of LLM traces |
| Langfuse | Evaluation & observability | Open source | Tracing LLM apps in production |
| LangSmith | Evaluation & observability | Freemium | LangChain and LangGraph users |
| Braintrust | Evaluation & observability | Freemium | Systematic evals before shipping |
| W&B Weave | Evaluation & observability | Freemium | Teams already on Weights & Biases |
| Helicone | Evaluation & observability | Freemium | Quick cost and usage visibility |
| promptfoo | Evaluation & observability | Open source | Prompt regression tests in CI |