GPT-6 Is Already Hitting Near-100% Success in Robot Arms

What would happen if we put GPT-6 in a humanoid robot? What benchmarks should we use? And the answer is, we know the answer. There have been multiple cases where GPT-6 has already been benchmarked with an embodied robot. So one of my favourite examples is a company called RoboCurve that right after the GPT-6 release released a benchmark showing that GPT-6, if you plugged it into a robotic arm, and gave it a visual channel as well, so it can see what the robot arm is doing and has channels of actuation for manipulating the robot arm, is able to achieve near 100% completion rate on tasks like picking up cups or moving around blocks. Next few months, as a lower bound, we're going to see GPT-6, 5.1, or 6.1, maybe something like that, solving general-purpose humanoid robot tasks.