H3-metal – Native MiniMax-H3 inference for Apple Silicon

Show HN: Mcptoon – Token-efficient MCP CLI client

Chicken Scheme 6.0

As AI eats the web, the internet’s collective memory is disappearing

LFM2.5 2.6B model competitive with 4x larger models

Show HN: Scroll through all 43252003274489856000 Rubik's Cube states

Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots

The “mechanical miracle” that ruined Mark Twain’s life

Mark Zuckerberg attacks 'closed' AI rivals as Meta returns to open models

Recycle – Floppydisks

The UK's war on anonymity has come to America

Stowaway – Take the window seat on any plane or satellite overhead

Sonic Pi v5

How Claude marks AI-generated content

Rust SIMD on the GPU

What's the best programming language for coding agents?

Faster floating point math with Rust's new API

Confessions of a Long-Distance Sailor

Publishing Schematics Before “Open Source” Was a Word

Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

Squeak 6.1

Tail-call optimization in C is relatively recent (2025)

Hyperspace

Learning more about Claude's mathematical capabilities

Show HN: Ante, a coding agent in a single binary that runs offline

Humanising LLM Outputs Is Dumb

Choral: Choreographic Programming for Java

Exploiting System Management Mode with a very long interrupt

Mars Bar from 1991 found – and it's 20g bigger than today's

Parametron: 50s Japanese computer that uses neither transistors nor vacuum tubes