OpenEvals
Add reusable scoring functions to existing tests for prompts and LLM application outputs.
Published by LangChain
How to use it
Choose a deterministic scorer for checks such as matching structured output, or configure a model-graded evaluator for a clearly written rubric. Retain the reference output and scoring settings with each case.
Call evaluators from a local test suite such as pytest or Vitest. Use fixed examples and compare repeated judge runs before treating a narrow score difference as a regression.
- Compatibility
- Python, pytest, Vitest
- Evaluation areas
- Assertions, Model-graded scoring, CI integration
- Access
- Local evaluation libraries. Model-graded scorers need a configured judge.