./daily — ← back to field notes

The link sheet

One-line pointers to today's local-AI, model, hardware, and healthcare-AI news. No commentary — just the headline, the link, and where it came from.

Sunday, 2026-11-08·last updated 2026-11-08 14:11 EST·9/9 feeds ok

[ hardware ]9

• b12211 — llama.cpp
• 4x Radeon AI PRO R9700 + 2x RTX 3090 on one x16 slot, using a PCIe switch — r/LL
• b12210 — llama.cpp
• b12209 — llama.cpp
• b12208 — llama.cpp
• b12206 — llama.cpp
• Huggingface - WOW.. 2 ZeroGPU Spaces for all free accounts? — r/LL
• b12202 — llama.cpp
• ASRock Rack DMRD16-2L+ Spotted with 16 DDR5 Slots for Diamond Rapids — STH

[ new models ]10

• Rumor: Moonshot Kimi roadmap — r/LL
• tencent/Qwen3.8-27B-DSpine-b16 · Hugging Face — r/LL
• I ran Qwen3.5-0.8B in pure Assembly — r/LL
• I fine-tuned Qwen3.5-0.8B to turn plain English into bash. What I learnt — r/LL
• b12207 — llama.cpp
• 2x256gb M5U: current status — r/LL
• b12205: webgpu : add COUNT_EQUAL support (#31578) — llama.cpp
• b12203: ggml : switch to F32 src1 in the CPU backend (#31854) — llama.cpp
• Crowd-v1 - The collaborative LLM — r/LL
• What is best local LLMs for strategic reasoning? — r/LL

[ local ai ]12

• GPT-5.5 dataset across my used 25 billion tokens! — r/LL
• Motion graphics video made with Qwen3.8 Flash Next GSQ RCO IQ3_XXS on strata — r/LL
• Has anyone used tensortorrent? — r/LL
• "Does censorship exist in [country]?", according to various LLMs — r/LL
• Users have started buying RTX 5060 Ti 16GB graphics cards en masse for the sake of AI. — r/LL
• Proposal: uncensoring LoRAs — r/LL
• Learning RL From Scratch: The What, Why and How of Double DQNs — r/LL
• OpenClaw Retrospective — r/LL
• Anybody tried Apex 35B and Axion 35B? — r/LL
• mtmd: support MageVL-Cowork model by ngxson · Pull Request #31819 · ggml-org/llama.cpp — r/LL
• Saw this benchmark where small changes to coding tasks made a big difference — r/LL
• Anyone running local AI on 16GB vram? — r/LL

[ benchmarks & papers ]1

• Towards ecologically meaningful foundation models — NatMI

[ tools ]5

• Anyone tried this yet? — r/LL
• The Qwen Flash Next alibaba URL thing happened to me too just now — r/LL
• Artificial Analysis: multimodal decision model Drex Max ranks #1. Both achieving SOTA an… — r/LL
• b12204: ci: move runners to Ubuntu 26.04 (#31529) — llama.cpp
• Show HN: Uncracked City, detective game where you question LLM suspects that lie — HN

[ healthcare ai ]7

• STAT+: Why the latest heart drug failures should make us all more skeptical of AI hype — STAT
• Lowering blood pressure after stroke or with diabetes poses challenges, in research and… — STAT
• STAT+: BIO joins the legal fight against Medicare pilot that ties drug prices to peer co… — STAT
• STAT+: Failure of Novartis Lp(a) drug raises questions about cardiovascular strategy — STAT
• STAT+: Despite Novo drug’s failure, cardiologists say they will still pursue anti-inflam… — STAT
• What a Swedish Mammography Trial Shows About Medical AI — HN-MED
• Healthcare AI needs a different model-development flywheel — HN-HC