./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.

Wednesday, 2026-10-28·last updated 2026-10-28 11:57 EDT·9/9 feeds ok

[ hardware ]8

• b11963 — llama.cpp
• b11962 — llama.cpp
• Used Qwen 3.8 flash next on underspeced PC to Root and gut LG TV of spy and adware — r/LL
• b11960 — llama.cpp
• b11954 — llama.cpp
• b11953 — llama.cpp
• DeepSeek-V4.1-Flash-1017 at 41 tok/s on one 3090 — r/LL
• Microsoft Maia 300 Shown at OCP 2026 — STH

[ new models ]10

• b11965: llama : remove simple batch splitting for DSV4 (#31300) — llama.cpp
• NVIDIA Nemotron 3.5 Transcribe and Granary v2: Intelligent Multilingual Transcription Yo… — HF
• I took EmbeddingGemma-2 (577M) from Rank #26 to #1 on the MTEB Audio leaderboard (MAEB)… — r/LL
• v0.8.0 — llama.cpp
• Someone built a 32-bit RISC-V CPU directly inside Qwen's model weights so it can execute… — r/LL
• b11958: server : support tool calls in /v1/realtime (#31287) — llama.cpp
• b11957 — llama.cpp
• AA is getting ridiculous now. Zero open weights in the Most Attractive Quadrant. — r/LL
• b11952: vulkan: clamp A prefetch to end_k in coopmat1 int8 matmul (#30283) — llama.cpp
• Artificial Analysis Intelligence Index v5 (major update) — r/LL

[ local ai ]12

• Alternative to artificialanalysis for assessing latest models? — r/LL
• AMA Announcement: MOSS, the Team Behind MOSS-Image, MOSS-TTS, and More (Friday, Oct 30… — r/LL
• CUDA: fuse MoE gate/up matmuls and GLU in MMQ by anujj · Pull Request #31188 · ggml-org/… — r/LL
• Gemma 5 models soon ?? — r/LL
• Diffuspace/Clover-35B-A3B-Preview · Hugging Face — r/LL
• 4B 3D coding model update: much better one-shot results and visual quality — r/LL
• Dirt cheap GPUs coming next year — r/LL
• They released a tesseract successor (datalab-to/indra) — r/LL
• SK Hynix expects demand to outpace supply through 2030 — r/LL
• New audio model: LFM3-Audio-2B — r/LL
• The reason why 5090's are $10k now... — r/LL
• LiquidAI/LFM3-Omni-2B-GGUF — r/LL

[ benchmarks & papers ]2

• Guidelines for evaluating structure-based methods for protein–ligand binding prediction — NatMI
• Efficient equivariant framework for universal charge density prediction — NatMI

[ tools ]1

• Survival LLM: AI assistant and emergency library that runs on your own computer — HN

[ healthcare ai ]12

• Senate candidate El-Sayed wants to ‘grill’ RFK Jr., calls him the ‘most incompetent HHS… — STAT
• STAT+: Why every health care AI company must face Epic — STAT
• STAT+: Pharmalittle: We’re reading about 340B battles, new results for a Lilly weight lo… — STAT
• STAT+: Lilly’s triple-G drug beats Ozempic in head-to-head study — STAT
• What has actually caused the decline in drug overdose deaths? — STAT
• STAT+: Lilly says ‘triple-G’ drug beats Novo’s Ozempic in diabetes study — STAT
• Opinion: How pediatricians like me are helping families stay calm and informed about mea… — STAT
• Have you been impacted by Epic’s actions? — STAT
• Opinion: Freezing my eggs showed me just how anti-feminist the fertility industry can be — STAT
• STAT+: As Epic’s dominance grows, its iron grip is choking innovation in health technolo… — STAT
• STAT+: Missouri AG warns Lilly to stop requiring data from hospitals seeking 340B drug d… — STAT
• STAT+: After two deaths in autoimmune cell therapy trial, Artiva is shutting down — STAT