Langfuse alternatives for tracing and evaluating production agents
The main Langfuse alternatives in 2026 are Comet Opik and Arize Phoenix for teams that still want to self-host, LangSmith for teams built on LangChain, Braintrust for eval-first teams, Pydantic Logfire for full-stack Python observability, and Datadog or MLflow for teams already on those platforms.
Checked
Langfuse is an open-source platform for tracing, evals, prompt management and datasets, with an MIT core and a free self-hosted edition. ClickHouse acquired it on 16 January 2026 and said there are no planned changes to licensing. Teams look at alternatives when they want a managed eval-first workflow, want a vendor that matches their framework or existing telemetry, or would rather not operate a self-hosted install.
The options below differ mostly in licence and hosting model. Each fits a particular kind of team running agents in production.
Why teams look past Langfuse
Operating the open-source edition
The self-hosted edition is free, but someone has to run, upgrade and back it up. Teams without that capacity often compare Langfuse Cloud with other hosted platforms.
Enterprise features
Code under the ee/ directories carries a commercial licence, and Self-Hosted Enterprise is custom-priced on top of a ClickHouse commercial plan.
Workflow fit
Teams that run evals as the release gate, or that already standardise on LangChain, Datadog or Databricks, may prefer a platform built around that workflow.
Other open-source and self-hosted options
Comet Opik
Apache-2.0, with a self-hosted version on the same codebase as the cloud product, plus an agent and prompt optimizer. Pro Cloud is $19 a month for 100k spans.
Arize Phoenix
Built on OpenTelemetry and OpenInference, deployable with Docker or Helm, with datasets, experiments, prompt management and cost tracking. Its Elastic License 2.0 is source-available, not OSI open source.
MLflow 3
Apache-2.0 tracing, scorers, LLM judges and production monitoring, and the default for Databricks shops.
Hosted and commercial options
LangSmith
LangChain's platform, with native LangChain and LangGraph support, OTLP ingestion and LangSmith Engine (beta) for failure clustering. Plus is $39 a seat a month; self-hosting is Enterprise only.
Braintrust
Eval-first, with experiments, scorers and CI joined to production traces and its own trace database. Pro is $249 a month; self-hosting is Enterprise only.
Pydantic Logfire
OpenTelemetry-native observability for the whole app, from database and HTTP calls to model calls, paired with the open-source pydantic-evals. Team is $49 a month, and Enterprise can self-host via Helm.
Datadog Agent Observability
Agent traces correlated with APM, infrastructure and logs. Free to 40k LLM spans a month; Pro from $160 a month on annual terms.
How to choose
If self-hosting is the reason Langfuse was on the list, Opik and Phoenix are the nearest matches, with the licence as the main difference. If the reason to leave is operating effort, the comparison is Langfuse Cloud against LangSmith, Braintrust or Logfire on price for the team's real trace volume.
Motiontic's engineers are vendor-neutral on tracing and eval tools. They pick a backend against the client's hosting, data-residency and budget constraints, keep instrumentation portable where the stack allows, build the eval suite from the client's own production failures, and leave the traces, datasets and evals with the client's team.
Langfuse alternatives for production agents. Prices in USD, as of 1 November 2026
| Licence / hosting | Strength | Fits when | |
|---|---|---|---|
| LangSmith | Proprietary, SDKs MIT; self-host on Enterprise; Plus $39 a seat a month | Native LangChain and LangGraph support; Engine clusters failures | The agent runs on LangChain or LangGraph |
| Braintrust | Proprietary, SDK Apache-2.0; self-host on Enterprise; Pro $249 a month | Experiments, scorers and CI joined to production traces | Evals are the release gate |
| Comet Opik | Apache-2.0; self-host or cloud, Pro $19 a month for 100k spans | Full platform plus an agent and prompt optimizer | Self-hosting under a permissive OSI licence matters |
| Arize Phoenix | Elastic License 2.0 (source-available); Docker or Helm | Built on OpenTelemetry and OpenInference | The team wants OTel-native tracing it runs itself |
| Pydantic Logfire | SDK MIT; Team $49 a month; Enterprise self-host via Helm | Full-stack OpenTelemetry plus online evals | A Python team wants app and agent traces together |
| Datadog Agent Observability | Proprietary, inside Datadog; free to 40k LLM spans a month | Agent traces beside APM, infrastructure and logs | The organisation already runs on Datadog |
| MLflow 3 | Apache-2.0 | Tracing, scorers, LLM judges and production monitoring | The team runs on Databricks or already uses MLflow |
Common questions
Is Langfuse still free to self-host after the ClickHouse acquisition?
Yes. Langfuse's self-host pricing page says the open-source edition is free without limitations on core features, and ClickHouse said there are no planned changes to licensing. Enterprise and Government self-hosted editions are custom-priced.
What is the closest open-source alternative to Langfuse?
Comet Opik, which is Apache-2.0 and self-hosts on the same codebase as its cloud product. Arize Phoenix is close in features but uses the source-available Elastic License 2.0.
Is Arize Phoenix open source?
Phoenix is source-available under the Elastic License 2.0, which is not an OSI-approved open-source licence. Its instrumentation layer, OpenInference, is Apache-2.0.
Can a team move from Langfuse to another tool without re-instrumenting?
Largely, if the agent is instrumented with OpenTelemetry, since Langfuse, LangSmith and Braintrust all accept OTLP traces and Phoenix is built on OpenTelemetry. Attribute names can still differ, because the GenAI semantic conventions are at Development status.
Related
AI agent development
Agents that do real work for users, with the evals, traces and guardrails to keep them in production.
LangSmith vs Langfuse
Hosted platform from LangChain against the open-source, self-hostable platform now owned by ClickHouse.
Langfuse vs Arize Phoenix
Two self-hosted tracing and eval tools compared on licence, OpenTelemetry, evals and hosted options.
Langfuse vs Braintrust
The open-source, self-hostable platform against the eval-first hosted platform.
Agent observability
Step-by-step traces of what an agent did, why, and what it cost.
Running agents that have to hold up?
Describe the agent, the systems it touches and what it must never get wrong. The first call is with a lab engineer, who will say plainly whether it is ready for production, and whether Motiontic is the right team.