Local AI is having its moment! Below is the number of new GGUF models created each month over the past 8 months & insights from our HF internal agent (May is partial): - 176,000 total public GGUF models on HF - Two distinct regimes: Oct–Feb averaged ~5.1K new GGUF models/month. Then March–April jumped to ~9.2K/month — nearly double the previous rate. - March was the inflection point (+55% MoM) — likely driven by a wave of new open-weight model releases being quantized to GGUF. - April sustained the momentum at 9.7K, suggesting this isn't a one-off spike but a new baseline. - The GGUF ecosystem is accelerating — the community is quantizing models faster than ever, likely thanks to better tooling (llama.cpp improvements, automated quantization pipelines, and more models supporting GGUF natively). Let's go!

CHINA SEMI LOCALIZATION plays going vertical today: AMEC 688012, NAURA 002371, ACMR 688082, and lesser known back-end equip play Hwatsing 688120. Theme gaining momentum on govt support, CXMT IPO, and I can't see many obstacles in near term. MS adding Naura to APxJ focus list today. Notes on AMEC, China's Lam:

Learning about the pros and cons of local ai in realtime here benchmarking Glimmer. This will be the first benchmark I run where the cost is $0...but also my first benchmark that will take 1-3 days to complete 👀

ANDREW NG OPEN-SOURCED A LOCAL AI COWORKER OpenWorker works across files, apps, and integrations to deliver finished work instead of just chatting Repo:

Stateful signatures trade systemic risk for localized risk. The failure mode is confined to individual wallets, not the network. @n1ckler at @TheBitcoinConf 2026 on why that's the better trade-off to explore.

Areas of thunderstorms and heavy rain will bring the risk for localized urban/poor drainage flooding overnight into Monday. The greatest chance for flooding will be across urban areas where repetitive thunderstorm tracks increase the duration of heavy rainfall.

Super agents have arrived on the desktop. With local AI, designers and engineers can build, customize and run domain-specific agents using their own data and workflows. NVIDIA NemoClaw on DGX Station brings together Nemotron 3 Ultra, Omniverse libraries and OpenShell in an open agent stack. 🔗 #SIGGRAPH2026#