PrismML releases Ternary Bonsai 2 models in 2B, 4B and 9B sizes
Overview
PrismML releases the Ternary Bonsai 2 series in 2B, 4B and 9B sizes, built on Alibaba's Qwen 3.5 base model and using ternary weights of -1, 0 and +1.
The company says the models need about 0.43GB, 0.95GB and 2GB of storage respectively, according to IT之家.
PrismML says the 9B model cuts output tokens by about 40% on standard instruction and daily math tasks. It also claims up to 5x throughput and 3x energy efficiency compared with full-precision versions on hardware including the RTX 4090 and M5 Pro. These performance figures are the company's own claims in the report.
Written by AI from the articles below · updated Oct 16, 2:31 AM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- IThome · AINewsPrismML releases Ternary Bonsai 2 models at 2B, 4B and 9B sizes
AIPrismML releases the Ternary Bonsai 2 series in 2B, 4B and 9B sizes, built on Alibaba's Qwen 3.5 base and using ternary weights of -1, 0 and +1, with storage of about 0.43GB, 0.95GB and 2GB. The company says the models cut tokens by about 40% on standard instruction and daily math tasks for the 9B model, and reaches up to 5x throughput and 3x energy efficiency versus full-precision counterparts on hardware such as the RTX 4090 and M5 Pro.
Heat trend
Not enough continuous observations to show a trend yet.