Skip to content
AI IntelligenceOct 20, 2026AI Intelligence
Article

Meta details Prometheus supercluster with about 1M GPUs at full scale, used to train Muse models: Meta detailed Prometheus, a...

Gigawatt-scale supercluster in Ohio that will have about one million GPUs at full scale. It was used to train the company's Muse foundational models.

Frontier EditorialSource: Techmeme
01

Source Brief

Meta details Prometheus supercluster with about 1M GPUs at full scale, used to train Muse models: Meta detailed Prometheus, a gigawatt-scale supercluster in Ohio that will have about one million GPUs at full scale. It was used to train the company's Muse foundational models.