Meta says Prometheus trained Muse and is scaling toward 1 million GPUs

Meta said on Oct. 19 that the first phase of Prometheus, the cluster it calls gigawatt-scale in New Albany, Ohio, is complete and trained the Muse models, including Muse Wonder. Meta says the site now spans hundreds of thousands of GPUs and is being scaled toward about 1 million. It reported more than 85 percent goodput on Muse Wonder pre-training runs and more than 90 percent on reinforcement-learning runs.