grok 4.6 takes more than 4x as many steps as other frontier models on computer use tasks
it is incredibly great at agentic work but sucks at clicking stuff on screens. almost every click it will have to try 4+ different times until it finally gets the right coordinates
this really shows the product ingenuity from the cursor/xai team. grok bot could have been made a while back
it's just the innovation of having the permanently running cloud VMs and the grok bot interface
I thought that software engineering was solved but after having to do some production debugging this weekend I realized its definitely not. You can still be a 10x engineer
grok 4.6 takes more than 4x as many steps as other frontier models on computer use tasks
it is incredibly great at agentic work but sucks at clicking stuff on screens. almost every click it will have to try 4+ different times until it finally gets the right coordinates