PulseAugur
EN
LIVE 14:11:44

PrismML shrinks 27B LLM for local iPhone execution

PrismML has successfully compressed its Bonsai 27B large language model to a 3.9GB file size, enabling it to run locally on iPhones. This achievement maintains approximately 95% of the original model's performance, representing a significant advancement for on-device AI and data privacy, particularly for Japanese enterprises seeking secure, offline AI solutions. AI

IMPACT Enables more powerful AI applications to run directly on mobile devices, enhancing privacy and offline capabilities.

RANK_REASON The release is a compressed model for on-device execution, fitting the 'tool' category rather than a frontier release.

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

PrismML shrinks 27B LLM for local iPhone execution

COVERAGE [1]

  1. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    A 27-billion parameter LLM running locally on your phone? PrismML just shrunk its Bonsai 27B model to 3.9GB, making on-device execution a reality for iPhones. B

    A 27-billion parameter LLM running locally on your phone? PrismML just shrunk its Bonsai 27B model to 3.9GB, making on-device execution a reality for iPhones. By retaining up to 95% of full-precision performance, this 1-bit model is a massive win for data sovereignty. It’s a majo…