Spanish startup Multiverse Computing launches free AI model that outperforms Mistral

The European race for open artificial intelligence has gained a new unexpected participant. Spanish startup Multiverse Computing, previously known primarily in narrow circles of quantum computing, published on Hugging Face a free compressed version of its HyperNova 60B language model. And in doing so, made a bold claim: according to the company, their model surpasses in quality an analogous solution from French Mistral — one of the flagships of European AI.

Multiverse Computing is a company from San Sebastián, which the industry classifies as a "soonicorn" — a startup rapidly approaching a billion-dollar valuation. Initially, the company focused on quantum computing for the financial sector and optimization tasks, but in recent years has actively expanded its expertise in large language models. HyperNova 60B is the result of this strategic pivot, and its appearance in open access signals serious ambitions.

Technically interesting is the approach itself: this is not merely the publication of another large model, but the release of its compressed version. Model compression is one of the hottest directions in modern AI engineering. The essence is to preserve the quality of answers from a model with tens of billions of parameters, while radically reducing computational resource requirements. This means the model can run on less expensive hardware, generate answers faster, and be accessible to a wider circle of developers. Multiverse Computing apparently uses its own developments in quantum-inspired optimization algorithms to achieve a high degree of compression without critical loss of quality — though the company is still disclosing details of its methodology in measured doses.

The claim of superiority over Mistral is a move that is simultaneously bold and calculated. Mistral AI, founded by former researchers from Google DeepMind and Meta, has become a symbol of European ambitions in generative intelligence. Its models — in particular, Mistral Large and more compact variants — are considered among the best open solutions on the market. Challenging precisely this player means asserting oneself at the highest level. However, one should treat such claims with caution: without independent benchmarks and detailed comparison on diverse tasks, claims of superiority remain a marketing move. The developer community on Hugging Face will surely conduct their own tests in the coming weeks, and then the picture will become clearer.

The fact of publishing a model freely and in open access deserves separate attention. In 2026, the AI model market is increasingly divided into two camps: proprietary solutions from large corporations like OpenAI and Google, and an ecosystem of open models, where Meta competes with Llama, Mistral, and now Multiverse Computing. For developers, startups, and researchers with limited budgets, each new quality open model is an expansion of possibilities. A compressed model capable of running on accessible hardware is especially valuable for small teams that cannot afford clusters of hundreds of GPUs.

For the European AI ecosystem, the emergence of a strong Spanish player alongside French Mistral and German Aleph Alpha is a positive signal. For a long time, Europe was criticized for lagging behind the United States and China in the artificial intelligence race, and efforts focused on regulation rather than technology creation. Multiverse Computing shows that innovation can be born beyond familiar tech hubs — in the Basque San Sebastián, not only in San Francisco or Paris.

The key question now is whether HyperNova 60B will confirm its ambitions in practice. If independent tests show that the compressed model is indeed competitive with Mistral solutions, this could shift the balance of power in the open model segment and attract the attention of major investors to Multiverse Computing. In any case, the fact itself that a small European startup with roots in quantum computing throws down the gauntlet to recognized leaders speaks to the maturity and dynamism of the market. Competition in open AI is intensifying — and everyone benefits from that.