Read what SHIP measured: benchmarks and studies of coding agents, with the data published beside them.
Our SWE-in-a-team benchmark graded thirteen coding agents and models in an SDLC loop. An Open-Weight builder resolved every ticket at half the price.
