Both trace agents, run experiments on datasets, score production traffic and version prompts. Arize is the one you can run yourself for free with Phoenix, and is now owned by Dynatrace; Braintrust is an independent hosted platform centred on experiments, with an analysis agent and a gateway in preview. Where the software has to run usually decides it first.[3][4]
Arize gives you a free self-hosted tool, Phoenix, plus a hosted platform, AX, while Braintrust is a single hosted platform built around experiments and prompt iteration. Dynatrace completed its acquisition of Arize on 1 October 2026 and says Phoenix and AX continue; Braintrust remains independent. Arize's hosted tiers start lower, at $50 a month for AX Pro against $249 for Braintrust Pro, and they meter different units.
| Decision factor | Arize | Braintrust |
|---|---|---|
| Ownership | Dynatrace, acquisition completed 1 October 2026; Phoenix and AX continue | Independent; $80M Series B in February 2026 |
| Products | Phoenix (self-hosted) and Arize AX (hosted, or self-hosted on Enterprise) | One hosted platform, with the Loop analysis agent |
| Licence | Phoenix: Elastic License 2.0 (source-available); AX: commercial | Proprietary platform; SDKs and Autoevals are open source |
| Self-hosting | Phoenix free; full AX self-hosting on Enterprise, including air-gapped | Enterprise only, data plane only; Braintrust runs the control plane |
| Online evals | AX only; Phoenix docs point to AX for continuous production evals | Yes, sampled scoring rules on production logs |
| Run-time checks | Partial: AX guards run in your code via Guardrails AI | None; scoring and alerts run after logging |
| Free tier | AX Free: 25k spans a month, 15-day retention; Phoenix free to self-host | Starter: 1 GB processed data, 10K scores, 14-day retention |
| First paid tier | AX Pro: $50 a month, 50k spans | Pro: $249 a month, 5 GB, 50K scores |
| Capability | Arize | Braintrust | 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 | — | ✓ |
| Gateway or proxy in the request path | Partial | ✓ | — |
| Licence and deployment | |||
| Open source | Partial | Partial | Partial |
| Self-hosting | ✓ | Partial | Partial |
Phoenix, Arize's self-hosted tool, installs with pip, Docker or Kubernetes at no cost and covers tracing, datasets, experiments, LLM and code evaluators and prompt versioning. Its licence is the Elastic License 2.0: you can read, modify and self-host it but not offer it to others as a hosted service. Arize AX can also be self-hosted on Kubernetes, connected, semi-restricted or air-gapped, under a commercial licence on the Enterprise plan.
Braintrust is hosted. Its Enterprise plan lets you run the data plane in your own AWS, GCP or Azure account, but Braintrust always operates the control plane: the UI, authentication and metadata. No fully self-contained deployment is documented. If a review requires everything inside your network, Arize has two routes and Braintrust has none.
Hosted regions differ slightly: AX SaaS runs in US, EU or Canadian data regions, and Braintrust's SaaS offers US and EU data plane regions.
Braintrust starts from the experiment. You bring a dataset, a task and scorers, run variants and compare, then promote versioned prompts through environments. Loop reads results, writes scorers and runs evals, and since September 2026 Patterns and Topics group production behaviour for review; several of these are in public preview. Online scoring rules run scorers asynchronously on sampled production logs.
Arize splits that work across its products. Phoenix handles tracing, experiments and evaluators, and runs server-side evaluators on every Playground experiment. Continuous evaluation of production traces at span, trace or session level, with monitors and alerting, sits in AX. AX adds an agent trajectory graph of which agents call which, agent-as-a-judge on every plan since September 2026 and a Decision Judge for high-volume structured evals.
For human review, AX has annotation queues that accept spans, traces, sessions or dataset examples, and Phoenix supports annotation through its UI, client or API. Braintrust supports assigned reviewers, multiple reviewers, blind reviews and required scores, though Starter allows one human review score per project.
Both accept OpenTelemetry traces. Arize maintains OpenInference, the conventions several other tools read; Braintrust converts incoming OTLP model calls into its own LLM spans.
Dynatrace completed its acquisition of Arize on 1 October 2026. Its announcement says Phoenix and AX will continue and that Arize capabilities will be integrated into Dynatrace over time, so packaging may change. For Dynatrace customers that may be a reason to choose Arize; for others it is a reason to watch pricing over the next few quarters.
Braintrust is independent and raised an $80M Series B in February 2026. Its risk is of a different kind: much of its newer work, including the hosted gateway, Patterns, Topics and the thread view, is in public preview, and its gateway has no SLA. Phoenix's own chat completions proxy, added in August 2026, does not support tool calling and is not positioned for production traffic.
Neither blocks a live agent action. Arize AX documents guards built on Guardrails AI that run in your own code around model calls; Braintrust's scoring and alerts run after logging.
As of 20 October 2026, Arize AX Free is $0 for 25k spans a month with 15-day retention, and AX Pro is $50 a month for 50k spans, 10 GB of ingestion and 30-day retention; both include unlimited users. Braintrust Starter is free with 1 GB of processed data and 10K scores, and Pro is $249 a month with 5 GB and 50K scores, also with unlimited users. Enterprise is custom on both, and it is the only plan with self-hosting of AX or of Braintrust's data plane.
Retention differs too: AX Free keeps 15 days and AX Pro 30. Braintrust keeps 14 days on Starter and 30 on Pro, extendable to 180 days at $0.50 per GB a month.
Arize meters spans; Braintrust meters processed data and scores, so a run with many small spans and few scores prices differently on each. Check Arize's pricing and Braintrust'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, in development and in production, and can act on a failing score through runtime policies, approval holds and a kill switch for a run, an agent or a team. Braintrust scores after logging, and Arize's guards run in your own code around model calls; neither documents approval holds or policies on tool calls. Prefactor has no prompt playground or prompt registry and its platform is hosted, so for prompt iteration or free self-hosting, Braintrust or 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.