Today's gold star is AeroVironment. Today's white star is Micron.
Grok 4.8 is a 2.5 trillion parameter model on xAI's new C++ stack; training finishes this week and reinforcement learning starts
read on XElon Musk said Grok 4.8, a 2.5 trillion parameter model, will finish its main training run this week and then start reinforcement learning, the stage where the model is scored on tasks and adjusted. He said the run used a new C++ software stack. A parameter is a number the model learned; 2.5 trillion sits with the largest public models. Watch for a model card, a public demo, and whether Grok 4.8 shows up in the product this month.