./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.
[ hardware ]12
- b11913 — llama.cpp
- b11912 — llama.cpp
- GPU unlocks (FP32 accumulation): Why can it be unlocked on Turing (RTX 20), Ampere (RTX… — r/LL
- b11910 — llama.cpp
- b11909 — llama.cpp
- b11908 — llama.cpp
- b11907 — llama.cpp
- b11906 — llama.cpp
- b11905 — llama.cpp
- b11904 — llama.cpp
- ASRock Creator Radeon AI Pro R9700 Review: Workstation Inference, AMD Style — STH
- Have 8 GB of VRAM and 32 GB of RAM? You can run Qwen 3.8 Flash Next at up to 900 t/s PP… — r/LL
[ new models ]6
- Qwen investigation traces release — r/LL
- I now have definitive proofs that my Qwen Flash Next is trying to send something to a re… — r/LL
- b11903 — llama.cpp
- Qwen4-27B's 4-way HC shift? — r/LL
- 3090 died, looking for replacement card — r/LL
- LitPivot: Developing research ideas through dynamic contextualization within the literat… — HF
[ local ai ]12
- PSA: Qwen-3.8-FN is about to win the 2026 ARC-AGI-3 challenge — r/LL
- Update: My 4B coding model is getting better at generating 3D scenes — r/LL
- Interactive neuron steering playground for Qwen3.5-4B that runs locally in the browser o… — r/LL
- Have you gone fully local? — r/LL
- I built an engine that runs Qwen-Flash-Next at nearly 400t/s on cheap(ish) hardware. 3x… — r/LL
- Price 5080 10 days — r/LL
- Oh nvidia… —>> NVIDIA reportedly not accepting RTX 5090 orders from board partners, even… — r/LL
- Local beats cloud, Qwen 3.8 FN — r/LL
- KLD is not the best metric out there and there isn't a single best one. — r/LL
- Qwen3.8-Flash-Next tried to send diagnostic file to a remote location, while stating oth… — r/LL
- People who tried gpu-unlock on 3090s and such - does this make any difference on mixed-p… — r/LL
- Fast voice agents with ConvFill — r/LL
[ benchmarks & papers ]4
- Qwen3.8-Flash-Next: Swift 1.5 vs Unsloth on Strix Halo (Halogen, gufo, llama.cpp, UD-Q4_… — r/LL
- OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? — HN
- Guidelines for evaluating structure-based methods for protein–ligand binding prediction — NatMI
- Efficient equivariant framework for universal charge density prediction — NatMI
[ tools ]1
[ healthcare ai ]12
- STAT+: Roche gets a lung cancer win with its KRAS drug divarasib — STAT
- STAT+: Nonprofit Bon Secours Mercy Health strikes $5.7 billion deal for surgery center c… — STAT
- STAT+: HHS relaxes security, transparency rules for health record software used by hospi… — STAT
- STAT+: House lawmaker probes PBMs for ‘punitive’ audits of independent pharmacies — STAT
- STAT+: GSK sees its return to cancer start to pay off — STAT
- STAT+: Merck ADC drug cut risk of death by one-third in advanced endometrial cancer — STAT
- STAT+: Rising premiums, a sad Medicaid gathering, and a new fraud crackdown — STAT
- STAT+: Incyte pancreatic cancer medicine induces lasting tumor responses, company says a… — STAT
- STAT+: Five biotech updates to read after ESMO this weekend — STAT
- STAT+: Pharmalittle: We’re reading about a Roche lung cancer drug, a Bristol CAR-T trial… — STAT
- A virus that afflicts cows makes the jump to a human — STAT
- The 4 stages of women’s heart health: New framework for doctors and patients — STAT
all days → · 47 links · generated from 9 sources