FrontierSeptember 18, 2026via SiliconAngle
PrismML launches Bonsai 2 27B, a high-intelligence AI model so small it fits on consumer hardware
Why it matters
Model compression via ternary quantization is enabling frontier-class performance on edge hardware—expanding where practitioners can deploy capable AI without cloud dependency or specialized accelerators.
Key signals
- Bonsai 2 27B is a multimodal model based on Qwen 3.8 27B
- Uses ternary quantization (3-state compression) to reduce model size
- Original Qwen 3.8 27B: ~56GB; compressed version runs on consumer PCs and high-end mobile devices
- Second-generation release (Bonsai 2)
- Announced September 18, 2026
The hook
PrismML's Bonsai 2 27B runs on consumer laptops. Here's how ternary quantization got a 56GB model down to fit.
Prism ML Inc. announced Thursday the launch of Bonsai 2 27B, the second generation of its ultra-compact multimodal generative artificial intelligence small enough to fit on PCs and some high-end mobile devices. The company said it used ternary, which uses three parts, to scale down its Qwen3.8 27B-b…