Ternary Bonsai 2 27B MTP needs about 9.1 GB of memory at its main file (a 6.4 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 2 quants, from 6.4 GB to 7.7 GB. Longer context needs more: at 32K tokens the KV cache alone adds about 8.7 GB.
Ternary Bonsai 2 27B MTP – 27B dense model with end‑to‑end ternary weights and bundled MTP head
Ternary Bonsai 2 27B MTP is a 27B‑parameter model using ternary weights and a Qwen3.8‑27B hybrid‑attention backbone. It offers a 262K‑token context and is released under Apache‑2.0. The model is available in two GGUF packings with different file sizes and memory footprints.
Good at
- 262K‑token context
- End‑to‑end ternary weights
- Bundled MTP head for self‑speculation
Limits
- No vision support in the base GGUF
- Requires a ternary‑aware engine
- Memory estimates are approximate
Best for
- Large‑context reasoning
- On‑device inference on laptops or single GPUs
- Applications needing a self‑speculating head
Who it’s for. Developers and researchers who need a large‑context language model that runs on a single GPU or laptop and who can use a ternary‑aware inference engine. The model is suitable for those who want to experiment with bundled MTP heads and low‑bit quantization.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 6.4 GB | 9.1 GB | Too big | Get file → | |
| — | 7.7 GB | 10.3 GB | Too big | Get file → |
For an 8‑GB VRAM budget, use the 6.4‑GB GGUF file which requires approximately 9.1 GB of VRAM at 8 k context. If you have 10 GB or more, the 7.66‑GB file needs about 10.3 GB of VRAM. Both files use unknown quantization but are designed for llama.cpp with ternary kernels.
What GPU do you need for Ternary Bonsai 2 27B MTP?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: the best quant that fits is the main file (7.7 GB file, about 10.3 GB in use).
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Ternary Bonsai 2 27B MTP
How much VRAM does Ternary Bonsai 2 27B MTP need?
About 9.1 GB at its main file with an 8K context window. That is the 6.4 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Ternary Bonsai 2 27B MTP on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 9.1 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Ternary Bonsai 2 27B MTP GGUF?
It depends on the quant: file is 6.4 GB, file is 7.7 GB.
What context length does Ternary Bonsai 2 27B MTP support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.2 GB at 8K and 8.7 GB at 32K.
Where can I download Ternary Bonsai 2 27B MTP GGUF?
The files are on Hugging Face in the ManniX-ITA/Ternary-Bonsai-2-27B-MTP-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Ternary Bonsai 2 27B MTP use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Ternary Bonsai 2 27B MTP on a GPU with 8 GB VRAM?
The 6.4‑GB GGUF file requires about 9.1 GB of VRAM at 8 k context, so it will not fit on an 8 GB GPU. A larger GPU is needed.
Does the model support vision tasks?
The base GGUF does not include a vision projector; use the separate mmproj file for vision.
What license applies to this model?
The model is released under the Apache‑2.0 license.