Arize Phoenix vs Langfuse vs LangWatch vs OpenLLMetry
A side-by-side look at scores, pricing and features — with RECATOOLS' ASEAN-aware verdict for each.
Arize Phoenix
Free, self-hosted LLM tracing and evaluation
Visit
|
Langfuse
Open-source LLM observability, tracing and evals
Visit
|
LangWatch
OpenTelemetry-native LLMOps for tracing, evals and agent simulation
Visit
|
OpenLLMetry
Open-source OpenTelemetry tracing for LLM apps
Visit
|
|
|---|---|---|---|---|
| RECATOOLS Score | 7.8 / 10 | 8.4 / 10 | 7.5 / 10 | 7 / 10 |
| Capability | ||||
| Value for money | ||||
| Ease of use | ||||
| ASEAN readiness | ||||
| API quality | ||||
| Pricing | Free | Freemium | Freemium | Open Source |
| Free tier | Self-host Phoenix free: no licence fees, usage limits or feature gates. Arize AX Free cloud tier: 25k spans, 1 GB ingest a month, 15-day retention | Self-host free under MIT (enterprise modules need a key); cloud Hobby plan free with 50k units/month, 30 days data access, 2 users | Developer: free forever, 50k events/mo, 14-day data, 2 users, 3 scenarios/simulations/custom evals. Self-hosting free under Apache 2.0 | OpenLLMetry is free and open source (Apache-2.0); Traceloop's hosted free tier covers up to 50K spans a month with 24-hour retention |
| Paid from | Phoenix is free. Arize's managed AX Pro plan is $50/month | $29/month (Langfuse Cloud Core) | Growth €29 (US$34) per core seat/mo + €5 per 100k events beyond 200k | — |
| Has API | ✓ | ✓ | ✓ | ✓ |
| Open source | ✓ | ✓ | ✓ | ✓ |
| Free to use | ✓ | ✓ | ✓ | ✓ |
| Users | — | — | — | — |
| Founded | 2023 | 2023 | — | 2023 |
| Maker | Arize AI | ClickHouse, Inc. | Independent (LangWatch Reasoning Engine B.V.) | ServiceNow |
| Verdict | Arize Phoenix is one of the most complete no-cost options for finding out why an agent or RAG pipeline behaved as it did: OpenTelemetry tracing, evals, datasets, experiments and a prompt playground, self-hosted with no f... |
Langfuse covers the core of LLM observability: nested traces with cost and latency, prompt versioning, datasets, experiments and LLM-as-judge evaluators, through framework-agnostic SDKs and OpenTelemetry. The MIT core se... |
LangWatch has become a genuinely open-source LLMOps platform, with strong agent simulation, evaluation and gateway features and generous self-hosting terms. Its pricing advantage over rivals is argued in its own blog rat... |
OpenLLMetry is the sensible default if you already run OpenTelemetry: a pip install and one init line, and your LLM calls show up as standard spans in the backend you already use. Coverage of providers, vector databases... |
| Full review → | Full review → | Full review → | Full review → |
Comparisons cover up to 4 tools. Scores are RECATOOLS editorial assessments; verify current pricing on each vendor's site.