Library / SDK
Observability & Evaluation
DeepEval
1Unit-testing 2framework 3for 4evaluating 5LLM 6outputs
the six Ws · specification
W1
Who
Maintained by Confident AI.
W2
What
A pytest-like framework offering LLM evaluation metrics for testing model and RAG outputs.
W3
Where
A Python library run locally or in CI, with an optional cloud dashboard.
W4
When
Reach for it when writing assertions and regression tests over LLM outputs.
W5
Why
Brings software-style unit testing to LLM quality with ready-made metrics.
W6
With
Works with pytest, LangChain, LlamaIndex, and any LLM provider.
W7
Watch
Maintained by Confident AI; open-source (Apache-2.0) library run locally. Evaluation stays local unless you opt into the Confident AI cloud (open-core).
alternatives
works with