runlocal.devCheck My GPU →

Ternary Bonsai 4B

Apache 2.0

Mid-tier ternary model. 1.58-bit weights {-1, 0, +1} give laptop-class footprint with capability that beats dense peers at the same byte budget. MLX 2-bit packed only for now.

Provider

PrismML

Parameters

4B

Context

32.768K

Released

2026-04-17

VRAM Requirements by Quantization

MethodDisk SizeVRAM RequiredFits GPUs
1.58-bit (MLX)0.9 GB1.2 GB15 GPUs

Benchmark Scores

mmlu62.5%
humaneval48%

Scores are approximate and may vary by quantization level.

Compatible GPUs (15)

HuggingFace

PrismML/ternary-bonsai-4b

View on HF →