Multiverse Computing is a Spanish software company that builds technology to make large language models smaller, under the name CompactifAI, as well as quantum-inspired optimisation software. Its offering includes an inference API for compressed models and private deployments, so that customers can run models in their own cloud, on-premises or on edge devices. The company was founded in 2019. The compression aims at running models with less computing power.
The headquarters is in San Sebastián (Donostia), Spain, and the legal entity is Multiverse Computing S.L. The sources reviewed name the Spanish public body SETT among its backers and identify no majority shareholder or foreign parent company. The website does not say where its own inference service stores or processes data, so customers should have hosting locations and processors confirmed contractually. No open-source status is stated for the products. For a European organisation, the option to run models inside its own infrastructure is the main sovereignty consideration.