What is LangWatch?
Open-core tracing, testing and evaluation for LLM apps.
Who is LangWatch best for?
LangWatch is best for AI engineers who want agents tested against simulated users and traced in production.
Teams shipping AI agents
Simulated users hold multi-turn text and voice conversations with your agent, and a judge scores each run before it reaches customers.
Engineers who want evals in CI
The same scenario tests run on your machine and on every pull request, and offline evals run from a notebook or the web app.
Watching agents in production
OpenTelemetry tracing records every step, token and cost, and online evaluations flag drops in quality on live traffic.
Teams that must keep data in house
The open-source platform runs on your own servers, and Enterprise adds SSO, audit logs and hybrid or air-gapped setups.
How much does LangWatch cost?
LangWatch is free on its Developer plan with 50,000 events a month or as self-hosted open source, and its Growth plan costs $34 per core seat per month plus $6 for every 100,000 events past the first 200,000.
- 50,000 events a month, 14 days of data access
- 2 users
- 3 scenarios, 3 simulations and 3 custom evals
- Full platform on your own servers with Docker Compose or Helm
- Unlimited events, limited only by your hardware
- Community support on GitHub and Discord
- 200,000 events a month included, then $6 per 100,000
- 30-day retention, then $4 per GB kept longer
- Unlimited lite users, simulations, evals and prompts
Enterprise, in the cloud or self-hosted, is priced on request and adds SSO, role-based access, audit logs, SLAs and hybrid, on-prem or air-gapped setups. Every LLM call, tool call, retrieval, evaluation or simulation step counts as one event, so one user conversation usually uses several. Volume discounts apply above 20 users. Prices are in US dollars; the page can also show euro prices (€29 per core seat). Full pricing on langwatch.ai. We checked these prices on .
How do readers rate LangWatch?
No one has rated LangWatch yet. Be the first.
Sign in to rate. One rating per account.
Is LangWatch open source?
Yes. LangWatch's code is public under the Apache 2.0 licence, so you can run it on your own server.
On GitHub
5,030stars
+85 (+1.7%) in the last 30 days
- Last commit
- 14 Nov
- Repository age
- 3 years
- Latest release
- 5 Nov
- Licence
- Apache 2.0
- Self-hosted
- Yes
- Repository
- langwatch/langwatch
What are the best LangWatch alternatives?
LangWatch is an open source alternative to Snowglobe, Okareo and Orq.ai. Helicone, Future AGI and Langfuse do the same job and are open source too.
SnowglobeFreemiumGuardrails AI's simulator that tests chatbots against realistic synthetic users before launch.
OkareoFreemiumSimulated users that test voice and text agents in CI.
Orq.aiFreeAgents built and evaluated with real-time tracing.
HeliconeOpen SourceLLM usage routed, debugged and analysed.
Future AGIOpen SourceTracing, evaluations and guardrails for production agents.
LangfuseOpen SourceOpen-source tracing, prompt management and evals for LLM apps.
Questions about LangWatch
What is LangWatch?
Open-core tracing, testing and evaluation for LLM apps.
Who is LangWatch for?
LangWatch is best for AI engineers who want agents tested against simulated users and traced in production.
Can I self-host LangWatch for free?
Yes. LangWatch is open source under Apache 2.0 and runs on your own servers with Docker Compose or a Helm chart, with unlimited events limited only by your hardware. The paid self-hosted Enterprise plan adds SSO, role-based access, SLAs and support.
How much does LangWatch cost?
LangWatch's paid plans are Growth at $34 per core seat per month.
Is LangWatch open source?
Yes. LangWatch's code is public under the Apache 2.0 licence, so you can run it on your own server.
What are the best LangWatch alternatives?
The closest alternatives on Recaply are Snowglobe, Okareo, Orq.ai, Helicone, Future AGI and Langfuse.
Where can I try LangWatch?
On its official website, langwatch.ai.
What counts as an event in LangWatch?
In LangWatch, every LLM call, tool call, retrieval, evaluation or simulation step is one event. A single user interaction with an agent usually creates several, so Growth's 200,000 included events cover fewer conversations than the number suggests.
Autosana
GTmetrix
Yaak
Chapar
Voiden
Bruno