MUSK’S AI IS SECRETLY RUNNING INSIDE SPACEX AND TESLA RIGHT NOW WITH 1.5 TRILLION PARAMETERS AND A 2 TRILLION MODEL COMES IN AUGUST Elon Musk confirmed on June 28 that Grok 4.5, xAI’s newest model built on a 1.5 trillion parameter foundation, has finished training and is now running in private beta at both SpaceX and Tesla. The company’s internal evaluations claim the model is comparable to or better than Anthropic’s Claude Opus, though no independent benchmarks have yet verified that claim. The model was originally targeted for late May 2026, so it arrived running about a month behind schedule. The announcement came alongside something more significant than a single model release. Musk outlined a plan to release entirely new AI models, trained from scratch, on a monthly cadence through December 2026. A 2 trillion parameter model is already in final training stages and expected to be complete by late July. Powering this development pace is xAI’s Colossus supercluster in Memphis, Tennessee, which now houses over 200,000 NVIDIA GPUs. The monthly release strategy is without precedent at this scale. Frontier AI training runs typically take months. Running them in parallel requires sustained compute investment that very few organizations in the world can currently match. Whether xAI can hold to this schedule remains to be seen. But if it does, the competitive landscape by December will look nothing like it does today, and every major lab will have had to respond to a new model every thirty days. Keywords: Grok 4.5 private beta, xAI monthly model releases, Elon Musk AI Colossus, SpaceX Tesla AI development