Both trace agents and score production traffic, but Arize, now owned by Dynatrace, starts from the eval workflow and Raindrop starts from live traffic. Arize gives you datasets, experiments, judges and prompt versioning, with a free self-hosted tool; Raindrop is a hosted monitor built to find recurring failures in production and compare changes on real users.[3][4]
Arize is a full evaluation suite with a free self-hosted tool (Phoenix) and a paid platform (AX), while Raindrop is a hosted monitor that classifies live agent traffic and groups recurring failures into issues. Dynatrace completed its acquisition of Arize on 1 October 2026; Raindrop is independent and announced a Series A in September 2026. Arize covers offline experiments and prompt versioning; Raindrop has neither as a hosted feature and leans on a local open-source tool for writing evals.
| Decision factor | Arize | Raindrop |
|---|---|---|
| Ownership | Acquired by Dynatrace, completed 1 October 2026; Phoenix and AX continue | Independent; Series A led by CRV in September 2026, $50M raised in total |
| Products | Phoenix (self-hosted) and Arize AX (hosted, or self-hosted on Enterprise) | Raindrop Cloud (hosted), plus Workshop, an MIT-licensed local debugger |
| Offline evals | Yes: datasets, experiments, LLM and code evaluators in both products | Partial: no hosted runner; Workshop runs evals locally, Simulations in early access |
| Production scoring | AX tasks run evaluators on live traces; Phoenix points to AX for this | Signals classify every new event; Issue Detection groups recurring failures |
| Judges | LLM judges, agent-as-a-judge and code evaluators you define | Signals drafted from a plain-language description; no custom judge prompt or model |
| Prompt tooling | Versioning and playground in both products | None documented |
| Self-hosting | Phoenix free; AX on Enterprise, including air-gapped | Beta: Enterprise VPC deployment for select partners |
| Entry pricing | AX Free: 25k spans a month; AX Pro: $50 a month | Hobby: 1,000 events a month; Pro: $299 a month plus usage |
| Capability | Arize | Raindrop | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | ✓ | ✓ | ✓ |
| Cost tracking | ✓ | ✓ | ✓ |
| OpenTelemetry support | ✓ | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | ✓ | Partial | ✓ |
| Online evals on production traffic | ✓ | ✓ | ✓ |
| LLM-as-a-judge scoring | ✓ | Partial | ✓ |
| Code-based evaluators | ✓ | Partial | ✓ |
| Human review and annotation | ✓ | Partial | ✓ |
| Prompt versioning and playground | ✓ | — | — |
| Act on findings | |||
| Runtime controls on live agent actions | Partial | — | ✓ |
| Gateway or proxy in the request path | Partial | — | — |
| Licence and deployment | |||
| Open source | Partial | Partial | Partial |
| Self-hosting | ✓ | Partial | 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. For a Dynatrace customer that integration is a point in Arize's favour. For everyone else it means packaging and pricing may move over the next few quarters.
Raindrop is independent. It announced a Series A led by CRV on 17 September 2026, bringing total funding to $50M, and in the same month launched Simulations in early access. It is a younger product with a narrower scope, and most of its framework integrations are still labelled beta.
This is the decision that separates the two. Arize is built for the full loop: you build datasets, run experiments, write LLM-as-a-judge or code evaluators, version prompts and then score production traffic with the same evaluators. Phoenix covers the development half for free; continuous production evals, monitors and alerting are AX features.
Raindrop assumes you already have an agent in production. You describe a behaviour in plain language, Raindrop drafts a classifier (a Signal) that runs on every new event and backfills a sampled 90 days, and Issue Detection groups recurring failures. Experiments then compare signal rates between cohorts of live traffic, so you can see whether a change helped real users.
The trade-off is control over judging. Arize lets you choose the judge prompt, the model and the code. Raindrop's classifiers are built by its own models, and you cannot bring your own judge. Raindrop's answer to pre-release testing is Workshop, an open-source local tool where a coding agent turns failures into evals and reruns your agent until assertions pass, plus Simulations for pull requests, which is early access.
On the production side the overlap is closer than it first looks. AX has monitors, alerting and automated issue detection of its own. Raindrop adds Slack alerts and a Triage Agent that investigates issues in the web app, in Slack or over MCP, with message limits per plan. If your team already lives in production dashboards, try both on the same week of traffic and compare which surfaces the failures you care about first.
Phoenix is free to self-host with pip, Docker or Kubernetes. Its licence is the Elastic License 2.0: you can read, modify and run it internally, but you cannot offer it to others as a hosted service. AX can be self-hosted on Kubernetes, including air-gapped, under the Enterprise plan.
Raindrop is hosted. Self-hosting is in beta as an Enterprise VPC deployment rolling out to select partners. Workshop runs on a developer's machine but is a debugger, not a copy of the platform.
Both accept OpenTelemetry. Arize maintains the OpenInference conventions that several other tools read. Raindrop accepts OTLP traces over HTTP with gen_ai attributes; it does not accept gRPC, metrics or logs. Arize AX also documents guards built on Guardrails AI that run in your own code around model calls; Raindrop documents no run-time checks on agent actions.
As of 20 October 2026, AX Free is $0 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. Phoenix costs nothing beyond the infrastructure you run. Raindrop's Hobby plan is free for 1,000 events a month, after which ingestion stops for the rest of the month, with one custom signal and 14-day retention. Pro is $299 a month plus $0.003 per event up to 1M events, then $0.002, with ten custom signals and unlimited retention.
Spans and events are different units, so model your own traffic before comparing. Check Arize's pricing and Raindrop's plans before you decide.
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, in development and production, and acts on a failing score through runtime policies, approval holds and a kill switch for a run, an agent or a team. Arize AX's guards run in your code around model calls and Raindrop alerts after the fact; neither documents approvals or policies on agent tool calls. Prefactor has no prompt playground or prompt registry and its platform is hosted, so if free self-hosting or prompt iteration is the need, 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.