PrismML Expands AI Models for Wearable Devices
News related to:PrismML · 2 min read
PASADENA, Calif., Oct. 15, 2026 /CourierPR/ -- PrismML, a U.S.-based artificial intelligence company focused on making AI more efficient and accessible, today announced the expansion of its Bonsai 2 family with new ternary models based on Qwen3.5. The new models are designed to support AI agents on everyday devices, such as personal computers, phones, smartglasses, and other wearables, while operating within resource-limited environments.
The new releases bring stronger capabilities to smaller deployment footprints, designed for the resource limits of personal computers, phones, smartglasses, and other wearables. The Ternary Bonsai 2 family now includes four sizes spanning 2B to 27B parameters, building on the first-generation Bonsai 1.7B, 4B, and 8B models.
Each of the new models has been trained using Google v5 TPUs and is designed for seamless integration with existing AI workflows. They are optimized for low-latency inference on consumer-grade CPUs and edge GPUs. The new models have compact footprints, with the 2B, 4B, and 9B models having footprints of 0.43GB, 0.95GB, and 2.00GB, respectively, more than 9 times smaller than their full-precision counterparts.
Across a suite of 19 benchmarks, the new Ternary Bonsai 2 models match or exceed the performance of their full precision counterparts on coding, math, and instruction following tasks. On tasks related to vision, agentic work, and tool-calling, as well as knowledge and reasoning, the models retain well over 90% of the benchmark scores of the full precision Qwen3.5 models. In all cases, the Ternary Bonsai 2 models show considerable performance gains over their first-generation Ternary Bonsai counterparts.
On hardware such as RTX 4090 and Macbook M5 Pro, the Bonsai 2 models deliver up to 3x the energy efficiency and 5x the throughput of their corresponding full-precision models. The Bonsai 2 9B model reaches up to 124 tokens per second (toks/s), while the Bonsai 2 2B model reaches up to 276 toks/s.
Their compact footprints are ideal for mobile deployments. On an iPhone 18 Pro Max, the Bonsai 2 9B model delivers up to 34 toks/s, while the Bonsai 2 2B model reaches up to 120 toks/s.
PrismML has demonstrated the new Bonsai 2 models in action, running a 1-bit Bonsai vision-language model on AI smart glasses powered by the Snapdragon AR1 Gen 1 Platform. This showcases the potential of the new models to operate within the memory and power constraints of wearable hardware.
The new Bonsai 2 models provide a foundation for bringing basic agentic functionality to future applications on phones, smart glasses, and other wearables. Developers can build assistants that interpret a user's request, reason over text or visual context, and use application tools to carry out short sequences of actions.
The new Ternary Bonsai 2 models are available for download under the Apache 2.0 license for free starting October 15, 2026. Developers, researchers, and other users can download the new Bonsai 2 models here.