Arize Phoenix vs LangSmith
Arize Phoenix
arize.com
LangSmith
langchain.comSide by side
| Figure | Arize Phoenix | LangSmith |
|---|---|---|
| Price from | Free | $39/mo |
| Free plan | Yes | Yes |
| Free trial | — | — |
| Open source | — | — |
| Web | Yes | Yes |
| Windows | — | Yes |
| macOS | — | Yes |
| Linux | — | Yes |
| API | Yes | Yes |
| Self-hosted | Yes | Yes |
| Plans published | 1 | 3 |
| Spec score · Llm Developer Tools | 8.2 (#7) | 9.0 (#3) |
| Spec score · Llm Monitoring | 8.1 (#12) | 8.9 (#6) |
| Spec score · AI LLM Evaluation Tools | 8.1 (#14) | 9.1 (#5) |
| Spec score · Prompt Engineering Tools | 8.1 (#12) | 9.1 (#4) |
| Spec score · AI LLM Observability Tools | 8.1 (#14) | 9.1 (#3) |
| Spec score · AI Agent Observability Tools | 8.0 (#19) | 9.1 (#4) |
| Spec score · LLM Observability Tools | 8.1 (#16) | 9.0 (#5) |
| Spec score · Model Monitoring Software | 8.1 (#16) | 9.0 (#4) |
| Deployment | Phoenix can run locally or be self-hosted using Docker, Kubernetes, or a cloud environment. | Deployment provides managed or self-hosted agent servers with streaming, cron scheduling, authentication, MCP support and more than 30 API endpoints. |
| Integrations | Documented integrations include frameworks and providers such as LlamaIndex, LangChain, DSPy, OpenAI, Bedrock, and Anthropic. | Documented integrations cover OpenAI, Anthropic, Bedrock, Gemini, Mistral, LiteLLM, OpenTelemetry, CrewAI, Vercel AI SDK, PydanticAI, AutoGen and many other frameworks and providers. |
| Prompt tools | Phoenix supports prompt versioning, side-by-side testing, and replaying calls to inspect how changes affect outputs. | Prompt Hub and Playground provide prompt versioning, testing, comparison and AI-assisted improvement. |
| Purpose | Phoenix helps teams trace, evaluate, debug, and improve AI applications. | LangSmith is a commercial agent engineering platform for building, testing, deploying, monitoring, and improving AI agents. |
| Support | The Phoenix documentation points users to a Slack community for questions and developer discussion. | The Developer plan includes community support, and the Enterprise plan lists a support SLA. |
| Tracing | It captures model calls, retrieval, tool use, and custom logic in traces that can be sent over OpenTelemetry. | Tracing records agent conversations and runs step by step, including prompts, model calls, tools, errors, latency and token usage. |
Plans
In their own words
Arize Phoenix
Arize Phoenix is free, open-source software for tracing, evaluating, debugging, and improving AI applications. It captures model calls, retrieval, tool use, and custom logic as OpenTelemetry traces. Teams can score traces and spans with LLM evaluators, code checks, or human labels. Prompt tools provide versioning, side-by-side testing, and call replay, while dataset experiments compare evaluation results across application versions. Documented integrations include LlamaIndex, LangChain, DSPy, OpenAI, Bedrock, and Anthropic, plus evaluation libraries such as Harbor, Ragas, Cleanlab, and MLflow. Phoenix runs locally or can be self-hosted with Docker, Kubernetes, or a cloud environment; Arize says self-hosting keeps traces in the user’s environment. Self-hosted authentication supports OAuth2 identity providers with an OpenID Connect well-known endpoint. The Phoenix plan costs nothing and leaves trace spans, ingestion, projects, and retention under user management. The project is ELv2 licensed and built by Arize AI and the open-source community.
LangSmith
LangSmith is a commercial agent-engineering platform for building, evaluating, deploying, monitoring, and improving AI agents. It supports offline evaluations on curated datasets and real-time scoring of production interactions. Evaluation methods include human review, heuristic checks, LLM-as-judge evaluators, pairwise comparisons, and custom Python or TypeScript evaluators. Annotation queues route runs to subject-matter experts for feedback that can calibrate evaluations, improve prompts, or augment datasets. The Playground compares prompt versions or model providers, and LangSmith chat can auto-improve prompts. Observability includes agent and retrieval tracing, token and general cost tracking, trajectory monitoring, online evaluations, and webhook or PagerDuty alerts. Traces can arrive through SDKs or an API from LangGraph, custom Python, or other stacks; SDKs are listed for Python, TypeScript, Go, and Java. Hosting ranges from cloud to hybrid and self-hosted Enterprise options. The Developer plan costs nothing, Plus costs $39 a month, and Enterprise pricing is custom.