Multiverse Computing Releases Free Compressed AI Model
▼ Summary
– Multiverse Computing addresses the large size and cost of AI models by creating compressed versions, such as its HyperNova 60B model, which is roughly half the size of its source model.
– Its compression technology, CompactifAI, is inspired by quantum computing and has been applied to OpenAI models, with the company open-sourcing more models in 2026.
– The compressed HyperNova 60B model offers lower memory usage and latency while maintaining potency, and an updated version better supports tool calling and agentic coding.
– Multiverse, a European company with global offices and enterprise clients, competes with firms like Mistral AI and positions itself as offering sovereign AI solutions.
– The startup is rumored to be raising a €500 million funding round at a valuation over €1.5 billion, following a $215 million Series B and support from Spanish regional governments.
The challenge of deploying powerful artificial intelligence often comes down to a simple issue of size and cost. Multiverse Computing, a Spanish startup, is tackling this barrier by releasing compressed AI models that maintain high performance while being far more practical for business use. Their approach centers on a proprietary compression technology called CompactifAI, which draws inspiration from quantum computing principles. The company has now made its latest model, HyperNova 60B 2602, freely available on the Hugging Face platform for developers.
This new release represents a significant step in making advanced AI more accessible. The HyperNova 60B model is approximately half the size of the OpenAI model it is derived from, requiring only 32GB of memory. This drastic reduction in footprint translates directly to lower operational costs, reduced latency, and easier deployment for enterprises. The updated version enhances capabilities in areas like tool calling and agentic coding, where efficient inference is critically important for controlling expenses.
Multiverse positions its technology as a competitive alternative to other leading models in the market. The company claims its HyperNova 60B outperforms Mistral AI’s Mistral Large 3 model in key benchmarks. Beyond this technological rivalry, the two European firms share several similarities. Both have expanded their operations internationally and serve a growing list of corporate clients. Multiverse counts major organizations like energy giant Iberdrola, engineering group Bosch, and the Bank of Canada among its enterprise customers.
The startup’s momentum appears to be attracting significant financial interest. Industry rumors suggest Multiverse is in the process of raising a new funding round of roughly €500 million, which would value the company at over €1.5 billion. While the company confirmed ongoing discussions with investors, it declined to comment on specific valuation figures or the size of the potential round. Reports also indicate the company’s annual recurring revenue reached €100 million at the start of the year, though Multiverse has not officially verified this number.
This growth trajectory, while still a fraction of industry leader OpenAI’s revenue, places Multiverse in a similar league to its European counterpart, Mistral. A key part of Multiverse’s strategy is emphasizing “sovereign solutions” within the AI sector, appealing to organizations and governments seeking alternatives to dominant U.S. technology providers. This positioning has already yielded strategic partnerships, including a recent collaboration with the regional government of Aragón in Spain.
Public sector support has been a consistent factor in the company’s development. The Spanish Agency for Technological Transformation participated in Multiverse’s $215 million Series B funding round last year. Since its founding, the startup has also received backing from the Basque regional government, which may soon see its investment yield the area’s first technology unicorn. By open-sourcing more of its compressed models in the coming years, Multiverse aims to further democratize access to powerful, efficient artificial intelligence.
(Source: TechCrunch)
