Spanish ‘soonicorn’ Multiverse Computing releases free compressed AI model

Spanish startup Multiverse Computing, rumored to be raising a significant funding round, is addressing the large size of LLMs with its compression technology called CompactifAI, inspired by quantum computing. The company has released a newer, free version of its HyperNova 60B model on Hugging Face, named HyperNova 60B 2602. This updated model is roughly half the size (32GB) of its source model (OpenAI’s gpt-oss-120B) but maintains high accuracy, offering lower memory usage and latency, and improved support for tool calling and agentic coding.

Multiverse positions itself as a European alternative to US tech giants, securing enterprise customers like Iberdrola and Bosch, and emphasizing the delivery of "sovereign solutions." The company is reportedly in discussions for a €500 million funding round at a valuation exceeding €1.5 billion, which, if confirmed, would solidify its status as a 'soonicorn.' Multiverse also benefits from regional and governmental support in Spain, including participation from the Spanish Agency for Technological Transformation (SETT) in its prior funding rounds.

(Source：TechCrunch)