Eval and tracing tools
Tools to store datasets, run experiments, trace production calls and review outputs.
Changes fast: check the latest updates too. Reviewed 17 Oct 2026.
Key ideas
- Open source: Langfuse, Arize Phoenix, promptfoo, Ragas, DeepEval, Inspect.
- Hosted: LangSmith, Braintrust, Weights & Biases Weave.
- Use OpenTelemetry GenAI conventions so traces are portable between tools.
- A simple pytest suite plus a spreadsheet of labelled traces is a valid starting point.