Gemini 4 Argon teaches Apollo humanoid cornhole from video

Google DeepMind's demonstration sits at the intersection of two threads: the launch of Gemini 4 Argon as a frontier model for complex engineering and scientific work, and the company's push to apply advanced AI reasoning to physical systems. The cornhole test is notable not just for the robot's seven consecutive shots, but for the pipeline behind it. Argon was shown raw video, then analyzed the environment, mapped spatial coordinates, and calculated whole-body joint movements before generating simulation code autonomously. That code became the training ground for Apptronik's Apollo humanoid.

The company frames this as a way to compress a simulation process that would normally require weeks of manual work into a few hours. If that holds, the bottleneck for teaching robots new physical skills shifts from engineering labor to the availability of video. DeepMind also reports Argon scored 87 percent on its Embodied Reasoning benchmark, which measures reasoning about physical and spatial problems in 3D environments. Alongside the robotics work, Google says it has already used the model for projects ranging from optimizing large software codebases to quantum computing, nuclear fusion and life-sciences research, positioning Argon as a general-purpose reasoning engine rather than a robotics-only tool.