Both trace agents, run experiments and score production traffic with LLM judges and code. Arize gives you a free self-hosted starting point in Phoenix and prompt versioning in both of its products; Fiddler adds real-time guardrail scoring and monitors classic ML models in the same platform, but has no prompt tooling. Arize has been part of Dynatrace since 1 October 2026.[3][4]
Arize offers a free self-hosted tool, Phoenix, plus a paid platform, AX, with prompt versioning in both; Fiddler is a single commercial platform that adds real-time guardrails and covers classic ML models alongside LLMs and agents. Dynatrace completed its acquisition of Arize on 1 October 2026, and Phoenix and AX continue. Fiddler is independent and has no prompt versioning or playground.
| Decision factor | Arize | Fiddler |
|---|---|---|
| Ownership | Acquired by Dynatrace, completed 1 October 2026; Phoenix and AX continue | Independent; $30M Series C in January 2026 |
| Scope | LLM and agent tracing and evaluation across Phoenix and AX | Classic ML models, LLM applications and agents in one platform |
| Licence | Phoenix: Elastic License 2.0 (source-available); AX: commercial | Proprietary platform; Python client is Apache-2.0 |
| Self-hosting | Phoenix free; AX on Enterprise, including air-gapped | Enterprise only: VPC, on-premises, or in your AWS account via SageMaker |
| Prompt tooling | Prompt versioning and a playground in both products | None |
| Run-time checks | Partial: AX guards run in your code via Guardrails AI | Partial: real-time guardrail scores; inline blocking through a third-party AI gateway |
| Free tier | AX Free: 25k spans a month, 15-day retention; Phoenix free to self-host | Guardrails-only free plan, by demo request |
| First paid tier | AX Pro: $50 a month, 50k spans | Developer: $0.002 per trace, hosted only |
| Capability | Arize | Fiddler | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | ✓ | ✓ | ✓ |
| Cost tracking | ✓ | Partial | ✓ |
| OpenTelemetry support | ✓ | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | ✓ | ✓ | ✓ |
| Online evals on production traffic | ✓ | ✓ | ✓ |
| LLM-as-a-judge scoring | ✓ | ✓ | ✓ |
| Code-based evaluators | ✓ | ✓ | ✓ |
| Human review and annotation | ✓ | Partial | ✓ |
| 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. Its announcement says Arize will keep supporting both Phoenix and AX and that Arize capabilities will be integrated into Dynatrace over time. Teams already on Dynatrace may see that as a reason to choose Arize. Teams that are not should expect packaging and pricing to move over the next few quarters and plan a large rollout with that in mind.
Fiddler is a venture-backed independent company. It raised a $30M Series C in January 2026, bringing its total funding to $100M, and has positioned itself since then as a control plane for AI agents. Its changelog shows steady releases through October 2026, including a hosted judge model, span annotations and new gateway integrations.
This is the clearest practical difference. Phoenix is free to self-host with pip, Docker or Kubernetes, and it includes tracing, datasets, experiments, LLM and code evaluators and prompt versioning. It uses the Elastic License 2.0, so you can run and modify it internally but cannot offer it to others as a hosted service. Continuous evaluation of production traffic, monitors and alerting sit in Arize AX, and Phoenix's own documentation points users there.
Fiddler has no free self-hosted version. Its Developer plan is hosted only, and VPC, on-premises and SageMaker deployments come with the Enterprise plan through sales. Its free plan covers guardrails only and is reached through a demo request. For a team that wants to try agent tracing this week without a sales conversation, Arize is the easier start. For an enterprise that will buy a deployment inside its own network either way, both offer one: AX self-hosting, including air-gapped, is also Enterprise only.
Arize treats prompts as first-class objects. AX has a Prompt Hub with versioning, diffing, rollback and environment tags, and Phoenix has a playground with version control and tagging. Fiddler has none of this; its Prompt Specs are templates for judge evaluators, not application prompts. If your team changes prompts weekly and wants each version tied to traces and experiments, Arize covers that and Fiddler does not.
Fiddler's strength is on the other side. Fiddler Guardrails score requests and responses in real time for toxicity and other harmful content, faithfulness, personal data and secrets, and your application decides whether to block. When Fiddler is plugged into Kong AI Gateway, AgentGateway, LiteLLM or Portkey, blocking or redaction can happen inline. Arize AX documents guards built on the Guardrails AI framework that run in your own code around model calls. Neither vendor documents approval steps or policy checks on an agent's tool calls.
Fiddler also monitors classic ML models for drift, performance, explainability and fairness in the same platform. If you run fraud or ranking models alongside agents and want one vendor for both, that weighs in Fiddler's favour.
Instrumentation moves fairly easily between them, because both accept OpenTelemetry. Arize maintains OpenInference, a set of conventions for LLM and agent spans, and AX draws an agent trajectory graph automatically for frameworks such as LangGraph, CrewAI and AG2. Arize ships Python and TypeScript SDKs, plus a Go client for Phoenix. Fiddler maps the OpenTelemetry GenAI conventions and offers native agent SDKs for LangGraph, LangChain, Strands Agents and Google ADK, all in Python; other languages send raw OTLP or use its REST API.
As of 20 October 2026, Arize AX Free costs nothing for 25k spans a month with 15-day retention and unlimited users, and AX Pro is $50 a month for 50k spans with 30-day retention. AX Enterprise is custom and is the only self-hosted AX option; Phoenix is free to run yourself. Fiddler's Developer plan charges $0.002 per trace, so 100,000 traces a month would cost $200 before any Enterprise terms. Note the units differ: Arize counts spans, of which one agent run can produce many, while Fiddler counts traces.
Check Arize's pricing and Fiddler's pricing against your own volumes; Arize's packaging may change as it moves into Dynatrace.
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. Of these two, Fiddler's guardrails come closest, but they check content rather than tool actions; Arize's guards run in your own code around model calls. Prefactor has no prompt playground or prompt registry and its platform is hosted, so if prompt versioning or free self-hosting is the requirement, Arize 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.