FrontierThe story, in brief
PrismML Releases Bonsai 27B: 1-bit and Ternary Builds of Qwen3.6-27B That Run on Laptops and Phones
5.9GB. That's all PrismML's Bonsai 27B needs to run on your laptop—quantized from Qwen's 27B model without capability loss.
The KeyNews take
Why it matters
Quantization breakthroughs that democratize frontier-class model inference are reshaping deployment economics. Running 27B models on consumer hardware changes which orgs can afford real-time AI applications.
The key facts
7 to knowBonsai 27B is 1-bit and ternary quantization of Qwen3.6-27B
Ternary variant: {−1, 0, +1} weights at 1.71 bits per weight
Ternary model footprint: 5.9GB ideal size
1-bit variant uses binary {−1, +1} weights
Both variants released under Apache 2.0
Targets laptop and mobile deployment
Architecture unchanged from source model
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: PrismML just released Bonsai 27B. It is a low-bit representation of Qwen3.6-27B, not a new pretrain. The architecture is unchanged. Two variants ship under Apache 2.0. Ternary Bonsai 27B uses {−1, 0, +1} weights at a true 1.71 bits per weight. Its ideal size is 5.9GB. 1-bit Bonsai 27B uses binary…