Yes

>>Multiverse Computing's Groundbreaking Free Compressed AI Model Poses New Challenge to Industry Leaders

Multiverse Computing's Groundbreaking Free Compressed AI Model Poses New Challenge to Industry Leaders

crypthub

about 8 hours ago

Free Release of HyperNova 60B

Multiverse Computing launched the compressed HyperNova 60B model on Hugging Face in March 2025, offering it at no cost. The 32 GB model is roughly half the size of OpenAI’s original 120B version while keeping comparable accuracy. By making a high‑performing LLM freely available, the startup aims to lower entry barriers for businesses worldwide.

Quantum‑Inspired Compression

The company’s CompactifAI technology applies quantum‑inspired algorithms to shrink model parameters and restructure architecture. Results show about 95 % of the source model’s accuracy with 50 % fewer resources, 45 % faster inference, and a 60 % reduction in memory usage. These gains enable real‑time tool calling and agentic coding at lower operational costs.

European Sovereignty and Competition

Positioned as a sovereign AI solution, Multiverse targets European firms wary of U.S. platforms. It claims superiority over France’s Mistral Large 3 in efficiency tests and has attracted backing from the Basque and Aragón governments. The startup is seeking a €500 million funding round that could value it above €1.5 billion.

Market Impact and Roadmap

The compact model opens advanced AI to SMEs, edge devices, and regulated sectors such as finance and healthcare. Multiverse plans additional open‑source releases in 2026, specialized industry models in late 2025, and multimodal compressed models by 2027. Analysts predict the AI compression market could reach $8.2 billion by 2028, driven by demand for cheaper, greener AI solutions.