Both trace multi-step agents, run experiments, score production traffic and version prompts. Arize suits teams that want a free self-hosted start in Phoenix and OpenTelemetry at the centre, or that already run Dynatrace; LangSmith suits teams on LangChain or LangGraph that want tracing, evals and deployment from one hosted vendor.[3][4]
Arize sells two products, Phoenix, a source-available tool you self-host for free, and AX, a commercial platform, while LangSmith is LangChain's single proprietary platform that you self-host only on an Enterprise contract. Dynatrace completed its acquisition of Arize on 1 October 2026 and says Phoenix and AX will continue; LangChain is independent. Both trace agents, run offline and online evals and version prompts.
| Decision factor | Arize | LangSmith |
|---|---|---|
| Ownership | Dynatrace, acquisition completed 1 October 2026; Phoenix and AX continue | LangChain, independent; $125M Series B in October 2025 |
| Products | Phoenix (self-hosted) and Arize AX (hosted, or self-hosted on Enterprise) | One platform covering tracing, evaluation, Engine, deployment and an LLM Gateway (beta) |
| Licence | Phoenix: Elastic License 2.0 (source-available); AX: proprietary | Proprietary platform; client SDKs are MIT |
| Free self-hosting | Yes, Phoenix | No; self-hosting is an Enterprise add-on |
| Online evals | AX only; Phoenix docs point to AX | Yes, automation rules with sampling, including multi-turn threads |
| Run-time controls | Partial: AX guards run in your code via Guardrails AI | Partial: LLM Gateway (beta) caps spend, restricts models and redacts model calls |
| Free hosted tier | AX Free: 25k spans a month, 15-day retention, unlimited users | Developer: 1 seat, 5k base traces a month, 14-day retention |
| First paid tier | AX Pro: $50 a month, 50k spans | Plus: $39 per seat a month, 10k base traces |
| Capability | Arize | LangSmith | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | ✓ | ✓ | ✓ |
| Cost tracking | ✓ | ✓ | ✓ |
| OpenTelemetry support | ✓ | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | ✓ | ✓ | ✓ |
| Online evals on production traffic | ✓ | ✓ | ✓ |
| LLM-as-a-judge scoring | ✓ | ✓ | ✓ |
| Code-based evaluators | ✓ | ✓ | ✓ |
| Human review and annotation | ✓ | ✓ | ✓ |
| Prompt versioning and playground | ✓ | ✓ | — |
| Act on findings | |||
| Runtime controls on live agent actions | Partial | Partial | ✓ |
| Gateway or proxy in the request path | Partial | — | — |
| Licence and deployment | |||
| Open source | Partial | Partial | Partial |
| Self-hosting | ✓ | ✓ | Partial |
Dynatrace completed its acquisition of Arize on 1 October 2026. The completion release says Arize will keep supporting both Phoenix and AX, and that Arize capabilities will be integrated into Dynatrace over time. Nothing has been withdrawn, but packaging and pricing may change as that integration happens.
LangChain, the company behind LangSmith, is privately held and raised a $125M Series B in October 2025. It moved self-serve customers to new plans this month, priced partly in LangChain Standard Units. If you already run Dynatrace, Arize's new home is a reason to look at it first. If you do not, the ownership change is a reason to check the AX contract terms before a long commitment.
This is the clearest difference. Phoenix is free to self-host with pip, Docker or Kubernetes, and includes tracing, datasets, experiments, LLM and code evaluators, annotation and prompt versioning. Its code is public under the Elastic License 2.0, which lets you modify and self-host it but forbids offering it to others as a hosted service. Continuous evaluation of production traffic, monitors and alerting are AX features, and Phoenix's documentation points users to AX for them.
LangSmith has no free self-hosted edition. Self-hosted LangSmith is an add-on to the Enterprise plan, deployed on Kubernetes with Helm or Terraform, and the BYOC option, where LangChain runs the control plane and you hold the data plane, is also Enterprise only. AX can also be self-hosted on Enterprise, including air-gapped. So for a team that cannot send traces to a vendor and has no budget for a contract, Phoenix is the only one of the three products that fits.
Running Phoenix yourself has a cost beyond the licence: you host it, upgrade it and keep its storage healthy. Teams that start on Phoenix and later need production monitoring usually move to AX rather than build those pieces themselves, and the same OpenInference instrumentation works for both.
Arize builds on OpenTelemetry and maintains OpenInference, the span conventions for LLM and agent work that its instrumentors emit. AX adds Agent Trajectory, a graph of which agents call which, set automatically for LangGraph, CrewAI, AG2 and others. Phoenix added DECISION spans for routers and classifiers in October 2026.
LangSmith traces through its SDK decorators, framework integrations and an OTLP endpoint, and it maps OpenInference and OpenLLMetry attributes on the way in. Its deepest integrations are with LangChain, LangGraph and Deep Agents, and its Trajectory view, opened to all users in September 2026, combines messages across a main agent and its subagents. Because both read OpenInference attributes, instrumentation written for one can usually be pointed at the other.
Neither platform holds or blocks a live agent action. AX documents guards built on Guardrails AI that run in your own code around model calls. LangSmith's LLM Gateway, in public beta since 30 July 2026, sits in the model-call path and can cap spend, restrict models and redact personal data. Neither covers tool calls; LangChain describes tool and MCP gateways as planned.
The two charge on different units. As of 20 October 2026, AX Free is $0 for 25k spans a month with 15-day retention, and AX Pro is $50 a month for 50k spans with 30-day retention; both include unlimited users. LangSmith's Developer tier is free for one seat and 5k base traces a month, and Plus is $39 per seat a month with 10k base traces; LangChain does not publish the overage rate per trace.
A larger team pays more on LangSmith as seats grow, while AX cost follows span volume. Phoenix carries no licence fee, only the infrastructure you run. Check Arize's pricing and LangSmith's pricing against your own volumes.
We sell the layer this section describes. Read it with that in mind.
Stated plainly, since Prefactor publishes this page: Prefactor is an agent evaluation platform that scores each step of a run and acts on what it finds through runtime policies on tool calls and data access, approval holds for risky actions, and a kill switch for a run, an agent or a team. Arize's guards run in your own code around model calls and LangSmith's gateway covers model calls only, so neither holds a tool action for a person today. Prefactor has no prompt playground or prompt registry, and its platform is hosted, so if free self-hosting is the requirement, Phoenix is the better fit. See the full comparison of 15 platforms.
Reviewed against public product and documentation pages on October 20, 2026. If a vendor has changed a feature, product name, or positioning since then, send a correction and we will update it. Numbered source links in the page body point to the ordered sources below.
Prefactor scores each step of every agent run and can hold or stop an action when a score says it should not happen. Start on the free tier: 5,000 spans a month, no credit card required.
Prefactor helps teams observe, evaluate, and improve their AI agents in production, across every framework and provider.