The open-source alternatives to LangSmith are Langfuse, Opik and Arize Phoenix. All three trace what an AI application did, evaluate its answers, manage prompts and keep datasets, and all three self-host. Langfuse has an MIT core and the largest following; Opik is Apache-2.0 across the whole repository and adds tests that run in continuous integration; Phoenix builds on OpenTelemetry across Python, JavaScript, Java and Go, under the Elastic License, which forbids reselling it as a service.
2 of 3 carry an OSI licence. 3 of 3 run on your own machines. Figures read from GitHub on September 17, 2026.
One project per need, with the reason from its own README and the one thing to know before you choose it.
Tracing of model calls, retrieval and agent actions, prompt management with versions, evaluations and datasets, self-hosted with Docker, Kubernetes or Terraform.
Know firstSome features sit in enterprise folders under a paid licence; ClickHouse now owns the project.
Agent tracing, LLM-as-a-judge evaluations, prompt management, production monitoring and a PyTest integration, all Apache-2.0 with no enterprise folders.
Know firstIts Docker Compose route is for development; production means Kubernetes.
Tracing built on OpenTelemetry with evaluations, datasets, experiments and prompt management, plus a built-in agent and an MCP server.
Know firstElastic License 2.0: you may not offer it to others as a hosted service, so this site labels it source available.
The facts that decide most choices. Each project's page carries the full panel.
| Project | Best for | Openness | Licence | Runs as | Own machines | Stars |
|---|---|---|---|---|---|---|
| Langfuse | The most widely used, MIT core | OSI | See LICENSE | Docker Compose, Kubernetes (Helm), Terraform, Hosted cloud | Yes | 34,744 |
| Opik | One Apache-2.0 licence, plus CI tests | OSI | Apache-2.0 | Docker Compose, Kubernetes (Helm), Hosted cloud | Yes | 22,104 |
| Arize Phoenix | OpenTelemetry across four languages | Source available | See LICENSE | pip, uvx, Docker, Helm | Yes | 11,523 |
Licence and star figures read from GitHub on September 17, 2026. Stars are shown as a dated fact and were not used to rank this page.
An open-core LLM engineering platform, MIT outside its ee folders and part of ClickHouse since January 2026, for tracing, prompt management, evaluations, datasets and a playground, with Python and TypeScript SDKs.
Tracing of model calls, retrieval and agent actions, prompt management with versions, evaluations and datasets, self-hosted with Docker, Kubernetes or Terraform.
Know firstSome features sit in enterprise folders under a paid licence; ClickHouse now owns the project.
Comet's Apache-2.0 LLM observability and evaluation platform, one licence across the whole repository, for agent tracing, LLM-as-a-judge evaluations, prompt management, production monitoring, CI tests and an MCP server.
Agent tracing, LLM-as-a-judge evaluations, prompt management, production monitoring and a PyTest integration, all Apache-2.0 with no enterprise folders.
Know firstIts Docker Compose route is for development; production means Kubernetes.
Arize's AI observability platform built on OpenTelemetry for tracing, evaluations, datasets, experiments and prompt management, under the Elastic License 2.0, which forbids offering it as a hosted service.
Tracing built on OpenTelemetry with evaluations, datasets, experiments and prompt management, plus a built-in agent and an MCP server.
Know firstElastic License 2.0: you may not offer it to others as a hosted service, so this site labels it source available.
The questions that settle it, in the order they usually come up.
Opik is Apache-2.0 throughout and Langfuse's core is MIT. Phoenix is Elastic License 2.0.
Phoenix is built on it and covers Python, JavaScript, Java and Go.
Opik's PyTest integration fails a build when answers get worse.
LangSmith, in the vendor's own words, is a "complete AI agent and LLM observability platform with tracing and real-time monitoring", by LangChain's own description. langchain.com. The cases below are where that still wins.
Stars, licence, last commit and last release are read from the GitHub API by a script every week and carry the date they were read. Nobody types them.
What each project does is written from its own README and LICENSE, read in full on the date shown, and phrased as the project's claim. The sources are listed at the end.
That a project is faster, better or cheaper than LangSmith. A price for anyone. A user count. A roadmap. If a fact is not on this page, it was not in the source.
Langfuse, for the most widely used option with an MIT core, tracing, prompt management and evaluations. Opik, for a single Apache-2.0 licence across the whole repository plus CI tests and production monitoring. Phoenix, for OpenTelemetry tracing across four languages, under the Elastic License.
Yes, all three. Langfuse lists integrations for OpenAI, LlamaIndex, Haystack, LiteLLM, the Vercel AI SDK and Mastra; Opik lists 64 integrations including CrewAI and Spring AI; Phoenix is built on OpenTelemetry and calls itself vendor, language and framework agnostic.
Opik's Apache-2.0 does, and Langfuse's MIT core does for the parts outside its enterprise folders. Phoenix's Elastic License 2.0 does not: it forbids providing the software to third parties as a hosted or managed service.
Yes. Each runs evaluations with a model as judge, for things like hallucination and retrieval quality, alongside code-based checks and human labelling. Opik can run them in continuous integration through PyTest, so a drop in quality fails a build.
Each starts with Docker or Docker Compose for development and moves to Kubernetes for production; Langfuse also ships Terraform templates for AWS, Azure and GCP. Phoenix is the lightest to try, with pip install arize-phoenix or a single uvx command.
Running one of these inside your own environment, with your own data and your own security rules, is the kind of work Reveneau does. Read how a forward deployed engagement works.