About Opik
Opik is an open source platform for tracing, evaluating and monitoring AI agents. It logs each step an agent takes so developers can find and fix errors from development through production.
What it does
Opik records the steps an agent takes, from user input and context retrieval to tool calls. Teams can run automated evaluations and LLM-as-a-judge scoring to measure quality. Experts can annotate traces to flag underperforming behavior. A diagnostics feature groups recurring issues and suggests likely fixes, even when no explicit error is raised. The platform also covers prompt management and production monitoring.
Key features
- Trace each step an agent takes, from retrieval to tool calls
- LLM-as-a-judge evaluations on traces
- Expert annotation of traces
- Recommended fixes for silent errors
- Prompt management and production monitoring
Who it's for
The platform suits ML engineers, agent developers and QA teams who want to test agents before release and watch them afterward. Teams with a single simple prompt may find the setup more than they need.
Pricing and good to know
The open source version is free to use. The vendor also offers a free tier and a paid enterprise platform for larger teams. Evaluations that use an LLM judge consume model tokens, so plan for that cost when you run large test sets.
Highlights
- Trace each step an agent takes, from retrieval to tool calls
- LLM-as-a-judge evaluations on traces
- Expert annotation of traces
- Recommended fixes for silent errors
- Prompt management and production monitoring
