PrismML’s new Ternary Qwen3.6 27B runs near fp16 precision on 10GB of memory!!!

Hey everyone, Tim from AnythingLLM and today PrismML dropped Bonsai 27B - which takes the same concept of BitNet /Ternary models the applied to the Bonsai 8B & Image models that can run on a phone with really good accuracy and performance and brought it to Qwen3.6 27B - which is actually an intelligent model. So we finally have a proper model beyond 8B that is using this new methodology! Bonsai 27B GGUF on M4 Pro via llama.cpp @32K inside OpenComputer. Prompt was \"Do browser research to build me a stylized