OLLAMA Bonsai 2 27B Uncensored is a local GGUF. ternary bonsai 2 27b. ollama. qwen 3.8 27b. lm studio. PrismML ternary kernels. 262K context. Apache-2.0. No safety checker.
- PrismML Ternary Bonsai 2 27B
- Qwen3.8-27B base
- PQ2_0 / PTQ1_0 quants
- dealignai CRACK GGUF, 0% HarmBench refuse
- lm studio notes
- Local GGUF, ~5.9-7.2 GB
- 262K context
- ternary kernels via PrismML llama.cpp fork
- ollama mirror row
- No safety checker
- Download the zip.
- Extract on Windows.
- Load the GGUF in the PrismML fork or lm studio.
- Stock Ollama cannot load ternary kernels. Notes in
files/notes/.
Ollama stock runtime? No ternary kernels. Use the PrismML llama.cpp fork.
Uncensored? dealignai Bonsai-2-27B-Ternary-CRACK-GGUF. 0% HarmBench refuse.
Windows 10/11. ~8 GB VRAM class or 32 GB RAM.
Apache-2.0 weights + MIT desk notes - Copyright (C) 2026 bonsai27buncensored
