Bonsai 2 27B: 5.9GB Ternary AI Model with 98.2% Capability Retention
PrismML’s Bonsai 2 27B compresses Qwen3.8 27B into a 5.9GB ternary model. Explore its architecture, benchmarks, 262K context, vision, tool calling, local setup, hardware support, and real-world trade-offs.