Spanish ‘soonicorn’ Multiverse Computing releases free compressed AI model | TechCrunch

Large language models have a problem: they are large. Multiverse computinga Spanish startup, is tackling this problem with compressed models that aim to close the gap between what frontier models can do and what companies can actually afford to deploy.

The secret sauce is CompactifAI, a compression technology inspired by quantum computing that the Basque company has applied to models released by OpenAI. Starting today, developers can access a newer version of Multiverse’s HyperNova 60B model for free on Hugging Face. The company also plans to open more compressed models in 2026 to support a wider range of use cases.

According to Multiverse, the models are smaller, but almost as powerful and accurate. At 32GB, the HyperNova 60B is about half the size of the model it’s derived from – OpenAI’s gpt-oss-120b – while boasting lower memory usage and lower latency. The updated version, called HyperNova 60B 2602also now better supports tool calling and agentic coding, where inference costs can be high.

One of the competitors that Multiverse claims to have defeated with HyperNova 60B is Mistral Large 3, one of the models released by the French decacorn Mistral AI. But besides the technological rivalry, the two European AI companies also have a lot in common.

Like Mistral, Multiverse has expanded beyond its home country, with offices in the United States, Canada and across Europe. Both companies also have business customers. In the case of Multiverse, Iberdrola, Bosch and the Bank of Canada are mentioned.

And while Multiverse isn’t officially a unicorn yet, it is now Rumor has it that a new funding round of €500 million will be raised at a valuation of over €1.5 billion. In a statement shared with TechCrunch, the company confirmed that active discussions with potential investors about a new funding round are underway, but added that it would be premature to comment on the valuation or the size of the funding at this stage. Multiverse also declined to comment on reports that its annual recurring revenue (ARR) hit €100 million in January.

If confirmed, this would still only be a fraction OpenAI’s $20 billion ARR; but not so far from Mistral, whose ARR increased to more than $400 millionpartly due to growing demand for alternatives to American technology. Multiverse positions itself in the same way in its latest press release as a company that can “deliver sovereign solutions across the entire AI stack.”

These geopolitical undertones have recently contributed to Multiverse’s security a collaboration with the regional government of Aragonin northeastern Spain. Spain’s Agency for Technological Transformation (SETT) also participated in the AI ​​startup’s $215 million Series B last year. Since its foundingMultiverse has also benefited from support from the Basque region – which could very soon count its first unicorn.