./daily — ← back to field notes

Daily links

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.

Thursday, 2026-10-15·last updated 2026-10-15 23:05 UTC·9/9 feeds ok

[ hardware ]12

• What a £300, 35W Snapdragon mini PC can actually do 👀: 35B coder at ~30 tok/s, and an 80… — r/LL
• b11669 — llama.cpp
• b11662 — llama.cpp
• mistralai/Timestral — r/LL
• b11658 — llama.cpp
• b11655 — llama.cpp
• b11647 — llama.cpp
• b11645 — llama.cpp
• b11643 — llama.cpp
• Chicken-coop shed AI PC — r/LL
• Very impressed with LFM2.5-24-A2B in Pi Agent — r/LL
• 4x R9700 builds with $127 used EPYCs: $10.7k all-in per machine. I built three (384GB VR… — r/LL

[ new models ]10

• b11671: model: add Muse Spark Open (text, vision and audio) (#30616) — llama.cpp
• Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint — HF
• Flux.3 Image Klein is now out — r/LL
• v0.7.0 — llama.cpp
• nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB · Hugging Face — r/LL
• Apertus 1.5 Release Notes — HF
• I tested Qwen3.8 27B KLD for 21 Models at 9 different KV quants that fit on 16GB VRAM (1… — r/LL
• b11644: kv-cache: honor PARTIAL_ONLY in the plain KV cache (#30514) — llama.cpp
• Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from M… — HF
• Welcome FLUX.3 Image Klein in Diffusers — HF

[ local ai ]12

• ibm-granite/granite-switch-4.1 30B/8B — r/LL
• Ternary Bonsai 2 (2B, 4B, and 9B) just released on Hugging Face. The 2B can run entirely… — r/LL
• FP4 QAD quant of Qwen3.8-Flash-Next from Nvidia — r/LL
• 128GB RTX Spark laptop for $3832. — r/LL
• Unsloth Muse Spark Open is out! — r/LL
• UNSLOTH MUSE SPARK OPEN!! — r/LL
• AtomicChat/Ling-3.1-flash-GGUF · Hugging Face — r/LL
• nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B model family released (base, NVFP4, BF16… — r/LL
• Add Nemotron Super 3.5 VL and Nemotron Bolt 74B A9B support by frozenblade1224 · Pull Re… — r/LL
• NVIDIA AI: So many models launching this week! And we have one of our own coming... — r/LL
• 137B A6.8B model from Microsoft? — r/LL
• Will there be a black-friday this year? NVIDIA RTX 5060 Ti 16 GB Surges Up To 144% In Ja… — r/LL

[ benchmarks & papers ]3

• Linus Torvalds's first LLM-assisted patch series — HN
• Why do LLM pelicans look like this? — HN
• An all-in-one foundation model for multimodal medical image restoration and enhancement — NatMI

[ tools ]1

• Introducing Muse Spark Open: the most capable US open-weight model to date — r/LL

[ healthcare ai ]12

• Trump administration advances plan for peptide drugs favored by RFK Jr. — STAT
• Merck and Moderna have ‘lot of exploring to do’ after initial cancer vaccine success — STAT
• The Trump administration is doubling down on health care fraud investigations — STAT
• Mass. lawmaker Auchincloss favors boosting U.S. biotech over targeting China — STAT
• STAT+: Former FDA commissioners Califf, Gottlieb, Sharpless lay into current HHS leaders… — STAT
• STAT+: Verastem’s KRAS inhibitor shows promise in pancreatic cancer — STAT
• After a melanoma diagnosis, mRNA cancer vaccine delivered a patient hope — and a path fo… — STAT
• STAT+: Pharmalittle: We’re reading about former FDA head calling DOGE a ‘disaster,’ an A… — STAT
• STAT+: ‘AI sandboxes’ are all the rage. Advocates scramble to control their future — STAT
• Kenya reports second Ebola case — STAT
• STAT+: On Wall Street, there’s trouble stirring for biotech — STAT
• Forty percent of high school students report binge eating in new CDC survey — STAT