Currently, it benchmarks backends in two ways.

1. Custom script to target specific cases
   1. Sequential requests (GET/POST)
   2. Multiple parallel requests (GET/POST)
   3. File upload (Multipart Requests)
   4. JSON parsing
2. Load testing by [k6]

Below are the results for both

Send one request at a time	Send multiple request at a time
Send files with multipart	Send 1.04MB of JSON to server and let it parse it

You can have a look at the config file on how it works.

A TL;DR would be it simulates the user increasing from 0 to 50, staying there for a minute then increasing to 100 and so on till it reaches 200 in step 50 and then decreases in step 100 till it reaches 0.

In the graphs below the Red line, represents the number of virtual/simulated users and the Blue line, represents the Average round trip time (ms) on the Y-axes and the time for test on X-axis.