Spanish startup Multiverse Computing released HyperNova 60B 2602 for free on Hugging Face, deliverin...
The AMW Read
Updates the player map with a high-performance efficiency specialist and validates the scaling-vs-compression debate by demonstrating superior reasoning at a fraction of the memory cost.
Spanish startup Multiverse Computing released HyperNova 60B 2602 for free on Hugging Face, delivering 1.9x better reasoning than Mistral 3 Large with 92% less memory usage. Their quantum-inspired CompactifAI technology achieves up to 95% compression while maintaining performance, cutting inference costs by 50-80%. This signals a market shift from raw model size to performance-per-dollar, with the AI model compression space projected to reach $8.2B by 2028. Following a $215M Series B and potential €500M raise at €1.5B valuation, Multiverse is positioning Europe as a leader in efficient, sovereign AI infrastructure. The free model strategy aims to attract developers and convert them into enterprise customers across energy, manufacturing, and financial services sectors.
