Opik
Tracing, evaluation and production monitoring for LLM apps, with the backend open-sourced as well
What is Opik?
A decorator on a function or a wrapper around a client is enough to start recording traces, and from there the same data feeds experiments that score responses automatically instead of by hand — hallucination, answer relevance, context recall and the rest. More than thirty integrations cover the usual frameworks and providers, and the whole platform can run locally or on Kubernetes rather than only as a hosted service. Its difficulty is not capability but overlap: it competes directly with two platforms already in this catalogue, so the choice turns on integrations and hosting rather than on what is missing.
What can you do with Opik?
- Start recording with one decorator — Adding a tracking decorator to a function, or wrapping a provider client, captures the call tree without restructuring the application, in either the Python or the TypeScript SDK.
- Stop reviewing every response by hand — Experiments run a dataset through the application and score the outputs automatically with metrics such as hallucination, answer relevance and context recall, turning review into a report rather than an afternoon.
- Attach to the framework you already use — More than thirty integrations cover LangChain, LlamaIndex, CrewAI and providers including Anthropic, AWS Bedrock and Google Gemini, so instrumentation is usually configuration rather than code.
- Run the whole thing yourself — The backend is open source too, with documented local and Kubernetes deployments — the traces and datasets do not have to leave your infrastructure to be useful.
- Keep prompts and configuration versioned — Prompts and agent configuration are tracked alongside the traces they produced, which is what makes a regression traceable to the change that caused it.
Before you choose Opik
- It overlaps almost completely with Langfuse and Phoenix, both already catalogued here; if one of those is already running, the case for a second platform is about integrations and hosting, not missing capability.
- Self-hosting means operating the platform's backend services rather than importing a library, so a local trial is a container deployment and a production one is a cluster to look after.
Star history
21 Aug to 30 Oct · +1.5k
Frequently asked questions
Is Opik free for commercial use?
Opik is released under the Apache-2.0 licence — OSI-approved open source, which permits commercial use.
How can Opik be deployed?
Opik is available as Self-hosted / Managed cloud / Runs locally.