The Ternary Bonsai 2 model series, featuring 2B, 4B, and 9B parameter versions, has been released on Hugging Face. The smallest 2B model is capable of running entirely within a web browser, achieving speeds of up to 500 tokens per second on an M4 Max chip using custom WebGPU kernels. This release offers a new option for local and browser-based AI applications. AI
IMPACT Provides a new, efficient model for local and browser-based AI applications.
RANK_REASON Release of a new open-source model series with performance details. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →