frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Bleached corals thrive within days when touching healthy neighbour

https://www.nature.com/articles/d41586-026-03266-x
77•alkyon•20h ago•6 comments

Gold is now the top U.S. export

https://www.numismaticnews.net/gold-now-top-u-s-export
60•iamnothere•2h ago•14 comments

Development statistics for the 7.2 kernel

https://lwn.net/Articles/1088776/
94•farlight•2h ago•59 comments

Source code of the Estonian online voting system

https://github.com/valimised/ivxv
46•theanonymousone•16h ago•13 comments

The Commodore 64 hardware simulator: break chips and pins, see the results

https://github.com/antirez/c64-hw-simulator
59•antirez•2d ago•11 comments

Furifont, a Japanese font that shows furigana

https://furifont.antfu.me/
129•plaguna•2d ago•24 comments

Making chicken transparent using Dorito dust

https://chillphysicsenjoyer.substack.com/p/making-chicken-transparent-using
189•crescit_eundo•5h ago•55 comments

DeepSeek-v4.1-Flash-1017

https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017
533•theykk•11h ago•489 comments

Dam removals are bringing fish back across the northeastern United States

https://www.smithsonianmag.com/science-nature/dam-removals-are-bringing-fish-back-across-the-nort...
33•sohkamyung•19h ago•3 comments

Pele – Mermaid diagrams as themeable SVG

https://pele.run/
55•dchest•3h ago•6 comments

Artie (YC S23) Is Hiring Enterprise AES and Senior Engineers

https://www.artie.com/careers
1•j-cheong•2h ago

tsc-ts: TypeScript written in TypeScript

https://github.com/vercel-labs/tsc-ts
31•tosh•2h ago•29 comments

OpenCode Hacked: User credentials leaked

https://twitter.com/opencode/status/2111516673953935833
220•pranshuchittora•1h ago•48 comments

Solar surpasses nuclear power globally for first time

https://www.pv-magazine.com/2026/09/29/solar-surpasses-nuclear-power-globally-for-first-time/
362•xbmcuser•5h ago•205 comments

Vintage Cassettes: 1963-2010

http://vintagecassettes.com/
24•austinallegro•2d ago•2 comments

Pattern Matching Beats Narrative Reasoning

https://blog.scottstevenson.net/p/pattern-matching-beats-narrative
8•tosh•1h ago•1 comments

Rendering pretty sky in Around The World game

https://frozenfractal.com/blog/2026/10/12/around-the-world-34-a-pretty-sky/
100•ibobev•1d ago•6 comments

AI Can Write the Code. Who Verifies It?

https://mihaisplace.com/blog/ai-can-write-the-code-who-verifies-it
10•mmarvramm•1h ago•15 comments

You could have invented transformers from memory

https://alexwang.ai/posts/associative-memory/
443•speckx•1d ago•104 comments

Show HN: Parin – A 2D game engine between raylib and Godot, written in Dlang

https://github.com/Kapendev/parin
12•Kapendev•1d ago•2 comments

Show HN: High-quality duck audio synthesizer

https://celphase.github.io/duck-pond/
34•celphase•6h ago•6 comments

A skeptic's first contact with Kubernetes (2024)

https://blog.davidv.dev/posts/first-contact-with-k8s/
36•pmoriarty•16h ago•2 comments

A Tudor life in the time of plague (2020)

https://www.historytoday.com/archive/history-matters/tudor-life-time-plague
39•pepys•3d ago•3 comments

Show HN: Esp32sim – Run ESP32 firmware in the browser

https://github.com/mikroverk/esp32sim
71•joakimeriksson•5h ago•9 comments

Espresso machine can run Doom

https://www.gamespot.com/articles/this-espresso-machine-can-run-doom/
13•speckx•1d ago•3 comments

Show HN: Llmspeed – feel how fast a local LLM runs before you buy the hardware

https://llmspeed.dev/
213•albinstman•15h ago•68 comments

The First Candid

https://www.newyorker.com/magazine/2026/10/19/the-first-candid
37•Thevet•1d ago•8 comments

Show HN: Things to Do API

https://thingstodoapi.com/
13•invalidmonk•19h ago•3 comments

The future of programming (1982)

https://dl.acm.org/doi/10.1145/358453.358459
23•vismit2000•8h ago•4 comments

Israeli settlers attack European diplomats in the West Bank

https://www.ft.com/content/35a08f82-6830-41fa-83ae-d3765f3970e0
17•michaelhillaert•22m ago•2 comments
Open in hackernews

Qwen3.8-27B on the iPhone 18 Neural Engine at 15–70 tok/s (1.5 bpw)

https://twitter.com/anemll/status/2111517143087067309
28•anemll•1h ago

Comments

spacecoder2•1h ago
Anyone know how hot an iPhone gets running the Neural Engine flat out for 10 minutes?
anemll•1h ago
It's warm but not hot. On M6 it's about 50% less energy consumption in my tests vs GPU. I think iPhone should be similar
GeekyBear•39m ago
The neural engine is optimized for energy efficiency, and the GPU is optimized for performance.

You're going to get much better battery life and much less heat running on the neural engine.

analog_daddy•45m ago
Hi. Could anyone share any recommendations (hopefully with actual links) for App Store apps to try running such local models? I’m not really interested in side-loading since it currently seems (to me) more of an hassle than it’s worth.

App Store search is broken in a sense that there might be so many apps claiming to be local but once you read reviews, it’s just another scam where it’s it’s usually sending your query to remote server.

simonw•42m ago
My current favorite is this one: https://apps.apple.com/us/app/locally-ai-by-lm-studio/id6741...

It's free, not annoying, and has a bunch of model options.

layer8•33m ago
Non-app-store link: https://locallyai.app/

However this uses the GPU, not the Neural Engine like the submission.

jdgoesmarching•31m ago
Note that the model discussed in this article isn’t available in this app running on the phone. Still, it has some good local options like Gemma 4 and can link to the desktop LM Studio app which is cool.
woadwarrior01•31m ago
I'm the dev behind the Private LLM app, which was the first on-device LLM app on the App Store. Don't download it yet. I'm currently finalizing a brand new v2 rewrite which can run models that the copycat app (locally) cannot run, like Bonsai 2 27B, Gemma 4 12B etc.

https://x.com/private_llm/status/2101769928252473488

Retr0id•38m ago
OP, asking for upvotes is against the HN guidelines: https://x.com/anemll/status/2111531629529456771
createful•20m ago
Not Qwen, Bonsai. Although it is based on Qwen, and achieves similar scores, it is not the "full" Qwen.

Last I remember, when trying to run a model above 1-2 GB on iOS, it would refuse to do so, at least from Locally AI. iOS only allows apps to use about 1-2 GB of RAM before being killed. Though using their Neural APIs might bypass this.