Tag archive
PrismML shrank a 27-billion-parameter model from 54 gigabytes to 3.9 by storing each weight as a single bit, and published the benchmark table showing tool calling and vision falling off a cliff.
Moebius is a small inpainting model claiming far-larger-model quality, and a developer ported it to run on your own machine in a browser tab -- no server, no upload.
Researchers distilled a large AI agent's memory skills onto a compact 4-billion-parameter model, raising its success rate on a household-task benchmark from 4% to 78% while running three times faster than its 72B teacher.
Samsung R&D UK and Queen Mary University of London published DuoMem, a distillation method that took a 4-billion-parameter agent from a 4.3 percent task-success rate to 77.9 percent, nearly matching a 72-billion-parameter teacher model's 87.1 percent
导语 2026 年 WWDC 上，Apple 花了 40 分钟讲一个话题：AI 如何在 iPhone 上本地运行，不联网、不传数据、不消耗云端算力。这不是 Apple...
Technical Analysis of Google's Gemma 4 12B: A Paradigm Shift in On-Device AI 1....
Anthropic's Persona Selection Model (PSM, 2026) makes the claim explicit: "A persona is not the...
The architecture in one paragraph. Twenty-five specialist brains. Six subsystems (Multi-Brain Orchestration, Agent Tooling, Knowledge and Memory, Artifacts
https://blog.neutrontech.ai/2026/05/28/slack-notion-and-google-workspace-have-a-new-problem-hubyn/ F...