A run now writes one HTML report that a browser draws and a build server parses. Every measurement is in the markup, so the file a pipeline reads is the file a person opens: plans with every step and check, a race with its actors on one clock, and a load test with throughput, latency and active users charted. Export writes it from a finished test plan or load test, and the command line writes it with --html.

The variables panel is the full height of the window and slides in and out, with the line between it and the request to drag. One control in the header says which combination of dimensions the requests run against, and opens a list to change any of them.

Response times read as measurements rather than as columns of digits. A plan and a load test both show one row per request, slowest first, each on a bar scaled to the slowest, so the one worth looking at is the one your eye lands on. A plan that ran once shows one time instead of repeating it three times. Export is a control beside the run button in both.

The example workspace varies along Env, Region and UserGroup, which is what an API client is usually pointed at, and every value of every dimension changes something you can see.