frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms

https://github.com/firelex/jeff
401•firelex•8h ago•154 comments

Phyllotaxis: An audio-reactive LED display

https://jagi.studio/posts/phyllotaxis/
23•evakhoury•12h ago•2 comments

1996 chat room simulator connected to Win95 and System 7 web desktops

https://lolchat.rip/
70•henrychannel•5h ago•39 comments

Pirating the Pirates

https://mubi.com/en/notebook/posts/pirating-the-pirates
480•piotrgrabowski•13h ago•242 comments

Show HN: Pac-Bench – How well can models one-shot a Pac-Man game?

https://jonclegg.github.io/pacman-bakeoff/
21•thefourthchime•6h ago•10 comments

Tank Body Problem

http://www.jimsitu.com
55•jimbooonooo•4h ago•9 comments

MicroLLM Lab – Try 7 tiny LLM's in the browser

https://stateofutopia.com/experiments/microllmlab/
185•logicallee•10h ago•70 comments

12,000-year-old Göbeklitepe burials explain scattered bones

https://archaeologymag.com/2026/09/gobeklitepe-burials-hundreds-of-scattered-bones/
111•yusufaytas•2d ago•27 comments

ESP32S3 cluster running 1.58-bit (BitNet) Language model

https://github.com/Low-Zi-Hong/ESP32s3-LLM-Cluster
62•nkko•7h ago•8 comments

California farmers are struggling to sell grapes as demand for wine drops

https://www.kqed.org/news/12101534/california-farmers-are-struggling-to-sell-grapes-as-demand-for...
124•randycupertino•9h ago•290 comments

Sonnet 5.5

https://www.anthropic.com/claude-sonnet-5-5
686•D2OQZG8l5BI1S06•11h ago•456 comments

Who Killed Paulina Borsook's Career?

https://www.wired.com/story/paulina-borsook-profile/
20•danielmorozoff•1d ago•3 comments

Simulating Airband Am Radios

https://bitbashing.io/am-radio.html
4•shellpipe•2d ago•0 comments

Scientists solve 1840s space weather mystery

https://arstechnica.com/science/2026/09/scientists-solve-1840s-space-weather-mystery/
90•gumby•9h ago•41 comments

Hijacking the PS5's RTMP stream

https://yashgarg.dev/posts/hijacking-ps5-rtmp-stream/
228•ibobev•13h ago•72 comments

World Labs Is Joining AMD

https://www.worldlabs.ai/blog/amd-announcement
239•mfiguiere•8h ago•98 comments

Kids turned low-traffic NPR Spotify comments into a secret group chat

https://www.thisamericanlife.org/897/transcript
344•simonpure•13h ago•198 comments

How to win a beer with high-dimensional statistics

https://jamiesimon.io/blog/how-to-win-a-beer-with-high-dimensional-statistics/
47•jamie-simon•2d ago•6 comments

Does Reddit have an astroturfing problem? What the data suggests

https://www.petervijeh.com/projects/reddit-astroturf
162•p-s-v•15h ago•202 comments

Updated Google Maps shows destruction of the city of Rafah

https://twitter.com/AliAbunimah/status/2103890594137309425
343•slowin•13h ago•186 comments

Nvidia wants to put a watchdog chip next to every AI agent

https://www.cnbc.com/2026/09/28/nvidia-releases.html
138•jonbaer•13h ago•162 comments

What is the best shape of a city? Modelling effect of urban form on distance

https://journals.sagepub.com/doi/10.1177/23998083261458842
27•rustoo•2d ago•13 comments

It's Time to Investigate the AI Labs

https://calnewport.com/its-time-to-investigate-the-ai-labs/
384•ibobev•9h ago•137 comments

The Art Forger Who Became a National Hero

https://priceonomics.com/the-art-forger-who-became-a-national-hero/
27•bookofjoe•2d ago•6 comments

What reversing, modernising old games tells us about the economic impact of AI

https://this.os.isfine.org/blog/posts/what-reverse-engineering-and-modernising-an-old-war-game-te...
96•keeda•2d ago•42 comments

Show HN: HN.watch – Videos of all Hacker News posts

https://hn.watch/
146•mrborgen•13h ago•87 comments

Cf: The Agentic CLI for the Cloudflare API

https://blog.cloudflare.com/cloudflare-cf-cli-launch/
140•macleos•13h ago•63 comments

Bluegraph – Explore NOAA buoy data, rebuilt in 3D from measured spectra

https://bluegraph.io/
15•polytap•3h ago•2 comments

Behold the pawpaw

https://www.cbc.ca/radio/thecurrent/pawpaw-tropical-fruit-canada-9.7356882
63•BiraIgnacio•1d ago•28 comments

Profit Margins of the Largest Companies

https://www.visualcapitalist.com/ranked-how-profitable-are-the-worlds-largest-companies/
18•teleforce•2h ago•6 comments
Open in hackernews

Show HN: Pac-Bench – How well can models one-shot a Pac-Man game?

https://jonclegg.github.io/pacman-bakeoff/
21•thefourthchime•6h ago
Benchmarks how well Harness+models can create a Pac-Man game from a single prompt:

“Create a Pac-Man game in a single HTML page”

Each model gets one shot — no follow-up prompts or fixes.

Comments

Computer0•6h ago
Opus 5-5 seemed like a perfect clone, with others displaying flaws in an initial look. Astra notably created a bunch of surrounding ugly crap to look at.
vunderba•6h ago
I was just coming here to say this. I looked at all of them, and Opus 5.5 at high effort seems significantly better than every other clone.

Buffered controls, different pathfinding AI for each ghost, level transitions, etc.

Gpt-5.6 sol high technically completed the assignment as well, but the gaps within the pipes used to construct the maze, the lack of pause when eating a ghost, etc made it feel far less polished.

strataspace•4h ago
I tried this with DOOM. Fable 5 did a pretty shit job. Astra made pretty crazy animated sprites and was pretty good considering.

The fact that these are at all playable and 100x my programming skill level is pretty depressing from a certain pov. The ThreeJS dude posted ab how demotivated he was to continue his work, and while I was never a dev that did much with webgl, I commiserate.

acomjean•57m ago
I wonder if we need better programming abstractions/ languages that can make programming easier for people.

It seems like a lot of the programming is cookie cutter type stuff (where these ai programs shine) and should be easier.

They can be amazing, but promoting “make a pac man game” seems like an exercise in which model has the best training code that was Pac-Man.

dang•1h ago
Recent and related:

https://news.ycombinator.com/item?id=49882889

nedo_var•24m ago
Curious if models grasp the ghost patterns or just react. Pac-Man's more complex than it seems for one-shot learning.
hoistway•19m ago
Always assumed Pac-Man was an easy solve for modern AI. Guess those ghost patterns are trickier than they look for one-shot learning.
_matthew_•13m ago
I don't think it makes sense to have the prompt be that short. This is basically a bench.ark of how models interpret an overly vague prompt. It should at least be "Create a pacman clone in a single html page. Make it faithful to the original" if that's what we're scoring it on.
blindflag•13m ago
I'm curious; can you explain why you picked Pac-Man, in particular?