September 18, 2026, (Inside AI) — PrismML has unveiled Bonsai 2 27B, a compressed large language model that runs on personal computers and high-end smartphones without cloud infrastructure. The model squeezes Alibaba's Qwen3.8 27B down to 5.9 gigabytes while preserving 98% of the original's benchmark performance, according to the company. The release signals that reasoning-capable AI no longer demands massive data centers for deployment.
The startup, founded by Caltech researchers and led by CEO Babak Hassibi, employs a technique called ternary weight compression. This method converts standard 16-bit model parameters into just three values: positive one, negative one, or zero. By reducing the information density at each weight position, the model file shrinks dramatically. The approach maintains reasoning capability because larger models contain redundant parameters that compression can eliminate without meaningful performance loss.
This marks a significant improvement from the original Bonsai, released in March, which retained 95% of benchmark performance. That model has been downloaded over 11 million times, according to PrismML, indicating substantial developer adoption. Additional smaller models have accumulated 2.6 million downloads, demonstrating market demand for device-based AI.
PrismML raised a $22.25 million seed round backed by Khosla Ventures, Cerberus Capital, and Caltech. Ion Stoica, Databricks co-founder and director of UC Berkeley's Sky Computing Lab, serves as an adviser. CEO Hassibi declined to comment on reports that PrismML is in talks with Apple, though the timing suggests potential device-side AI integration.
Read: Why AI’s Next Chapter Belongs to CPUs: The Shift to Efficient Inference
Running reasoning models locally addresses privacy concerns associated with sending user queries to cloud servers. Processing remains private because data never leaves the device, and operation costs nothing beyond the hardware already purchased. The company's next goal targets compressing hundreds-of-billions-parameter models within the next couple of months. Hassibi indicated that larger models offer more redundancy, making compression easier while maintaining intelligence. As model size increases, he suggested reaching 100% benchmark performance parity becomes achievable.
"Intelligence at your fingertips," Ion Stoica, Databricks co-founder and director of UC Berkeley's Sky Computing Lab, said, emphasizing that device-based AI transforms user experience without cloud dependency.
Bonsai 2 27B runs on NVIDIA GPUs via CUDA and on Apple devices (Mac, iPhone, iPad) via MLX, through custom low-bit kernels. Model weights are available today under the Apache 2.0 License. The development represents a fundamental shift from cloud-dependent AI toward distributed device-based reasoning.
