frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

How to turn compute into a financial asset

https://www.economist.com/finance-and-economics/2026/06/22/how-to-turn-compute-into-a-financial-a...
1•zczc•50s ago•0 comments

How sad should I be about ChatGPT? (2022)

https://robertheaton.com/chatgpt/
1•reasonableklout•1m ago•0 comments

Ask HN: In the age of agentic coding why no one talks about orchestration tools

2•ios-contractor•5m ago•0 comments

Causal Summit

https://causalsummit.com/
2•northlondoner•6m ago•0 comments

Don't waste Claude limits babysitting AI experiments

https://heyneo.com/claude-code
2•gauravvij137•11m ago•0 comments

Show HN: Treating PowerPoint decks like code

https://github.com/sirilsengolraj-source/presentation-skill
1•Sirilsengolraj•11m ago•0 comments

AI Coding Traps Every Engineering Team Should Know

https://jsdevspace.substack.com/p/the-8-ai-coding-traps-every-engineering
1•javatuts•13m ago•0 comments

Show HN: Multiserial – a serial terminal for macOS that does not feel ancient

https://github.com/tpp6me/serial-com-multiplatform
1•praveentp•14m ago•0 comments

Chesterton's Middle Finger

https://www.arp242.net/chestersons-finger.html
1•birdculture•17m ago•0 comments

What is the best coding harness as of June 2026?

1•alxh•17m ago•1 comments

Chattiness

https://www.lrb.co.uk/blog/2026/june/chattiness
1•Michelangelo11•18m ago•0 comments

System Prompts – Claude API Docs

https://platform.claude.com/docs/en/release-notes/system-prompts
1•ankitg12•18m ago•0 comments

Show HN: A simple web studio for generating videos with Seedance models

https://seadance-video.com/
1•Evanmo666•22m ago•1 comments

React Parallax

https://react-parallax.dev/
1•iampoul•25m ago•0 comments

Show HN: We matched full-context recall on ~1% of the tokens (open benchmark)

https://github.com/compresh/compresh-benchmarks/blob/main/epbench/WRITEUP.md
1•compresh•28m ago•0 comments

The AI Poet

https://www.johnderoulet.com/post/the-ai-poet
1•MrBusch•28m ago•0 comments

Elevated errors for Claude Opus 4.8

https://status.claude.com/incidents/2stpj60m2fr9
1•adamheath•29m ago•0 comments

A single API to manage Apple, Google, and Stripe subscriptions

https://crosspay.dev/blog/introducing-crosspay-the-future-of-cross-platform-monetization/
1•pcvetkovski•30m ago•0 comments

CoreGLP Denmark

https://www.facebook.com/CoreGLPDenmark.Get
1•konikathy•30m ago•0 comments

Visualization: Would 1M Starlink satellites look

https://static.tomaskafka.com/prototypes/1m-starlinks/
1•tomaskafka•31m ago•0 comments

Advice on Gifted Education

https://terrytao.wordpress.com/career-advice/advice-on-gifted-education/
1•pykello•31m ago•1 comments

Meta Invests $900M in CRED, Taps Founder to Head WhatsApp

https://www.bloomberg.com/news/articles/2026-06-22/meta-taps-new-whatsapp-boss-as-part-of-900-mil...
3•thisislife2•31m ago•0 comments

Horsewood Reviews Experts vs. Consumers – Who's Right?

https://finance.yahoo.com/sectors/healthcare/articles/horsewood-urgent-report-2026-horse-19110038...
1•kilzdaru•38m ago•0 comments

Everything*: An interactive voyage through all orders of magnitude

https://www.everything.one/
1•luu•39m ago•0 comments

GPT 5.6 Pro SVG output is INSANE

https://twitter.com/JaydenDavisNC/status/2069190974622822872
2•wahnfrieden•42m ago•0 comments

Show HN: Exeria Charts – A native, high-performance financial charting library

https://github.com/efixdata/exeria-charts
2•katomek•42m ago•0 comments

NASA's Ranger 7 descending on the Moon [video]

https://www.youtube.com/watch?v=nV9nK8dqqU4
2•JPLeRouzic•45m ago•0 comments

Sideways, Not Up: Why AI Is Not a New Abstraction Layer

https://www.sixsideddice.com/Blog/Thoughts/SidewaysNotUpWhyAIIsNotANewAbstractionLayer.html
1•rusk•45m ago•0 comments

Patch the Planet

https://openai.com/index/patch-the-planet/
1•jonbaer•46m ago•0 comments

EGC – your AI never re-explains your project when you switch sessions

https://dev.to/fmarzochi/i-got-tired-of-re-explaining-my-project-to-ai-every-session-so-i-built-e...
1•fmarzochi•48m ago•0 comments
Open in hackernews

Show HN: MinLlama – Llama 3.2 inference in ~100 lines of NumPy

https://github.com/timothygao8710/minLlama
1•timothygao•1h ago
I built minLlama because I wanted a Llama implementation that was easy to understand and hack for KV cache compression research. There is also a PyTorch and Jax version in ~140 lines.

Would be interested in feedback from people who have written transformer implementations before, are there any implementation "tricks" that I'm missing (e.g, cleaner KV cache for PyTorch/Jax or rope tricks)?