Soofi S 30B-A3B MoE Hybrid Mamba Transformer for DE/EN
July 9, 2026
Soofi S is a 30B parameter Mixture-of-Experts model using a hybrid Mamba-Transformer architecture. It activates only 3B parameters per token and maintains near-constant inference cache for high-throughput, long-context deployment in German and English.
HOW THIS AFFECTS YOU
●
builderYou can deploy high-throughput, long-context models with significantly lower memory overhead.
●
founderThis provides a competitive, sovereign open-source alternative for European-market applications.