Coding
Promptfoo saves models, prompts, examples and pass criteria as a reusable evaluation. Run it again after a change and inspect the differences case by case. Security testing can also generate and run probes against authorized applications.
For AI teams validating prompts, retrieval workflows or model upgrades before release. You need a model or application endpoint to test.
Community is free. External model providers may charge by usage. Enterprise and on-premise plans have custom pricing.
Compare output and pass status for two model configurations on the same inputs.
Choose providers and prompts, then define examples and assertions.
After testing an authorized application, review results grouped by risk category.
You still need a model or application, test inputs and pass criteria. Credentials and costs depend on your provider.
The official CLI supports CI/CD. Start with repeatable cases and choose thresholds based on observed failures.
Run them only in environments you control or have permission to test. Start with synthetic data and an isolated test service.
Community is free. External model providers may charge by usage. Enterprise and on-premise plans have custom pricing.
Enterprise and on-premise plans are custom-priced.
Compare models, prompts and AI application behavior with repeatable tests.
Visit website ↗Submit once. Get listed permanently after approval.
