PrismML released Bonsai 2 27B, a compressed version of Alibaba's Qwen3.8 27B that shrinks the model to 5.9 GB while matching 98% of its aggregate benchmark scores. Founded by Caltech researchers and led by Babak Hassibi, the startup uses ternary weights to reduce model size for PCs and smartphones. Backed by Khosla Ventures, Cerberus Capital, and Caltech, with Ion Stoica as adviser, PrismML plans to compress several-hundred-billion-parameter models next.
No score is assigned. Sources and their independence are shown in the citation chain below.