What's new in 26.2

Use the new Evaluate GenAI response step to validate the behavior and reliability of GenAI and LLM-based intelligent systems before production.

Configure the evaluation methods, metrics, and pass levels, providing confidence-based evaluation for non-deterministic responses.

For details, see Evaluate GenAI responses.

Note: This feature requires an Aviator license.

Run your manual tests autonomously, without writing scripts or converting tests to automated ones. The Autonomous Tester runs your tests unattended, reducing execution time, increasing test coverage, and accelerating release cycles.

Assign manual tests to the Autonomous Tester runner and set them to run automatically in a test suite. View detailed run evidence and results, including agent actions, step status, and recordings.

Use Aviator to analyze test and suite run results for faster issue resolution.

For details, see Autonomous Tester.

Note: This feature requires an Aviator license.

Create unit steps in the new Steps tab, as you would in a manual test. Adding steps lets you describe a unit's detailed actions without linking it to a separate manual test.

If a unit contains steps, these are reflected in the Manual Runner when you run the test, and in the Manual Run report as well. This enable you to more easily pinpoint MBT test failures.

If a unit contains steps, these are reflected in the Manual Runner when you run the test, and in the Manual Run report as well. This enables you to more easily pinpoint MBT test failures.

For details, see Units.