What will enter the public domain in 2026?
publicdomainreview.org
DeepSeek-v3.2: Pushing the frontier of open large language models [pdf]
huggingface.co
Beej's Guide to Learning Computer Science
beej.us
India orders smartphone makers to preload state-owned cyber safety app
reuters.com
Apple Releases Open Weights Video Model
starflow-v.github.io
Reverse math shows why hard problems are hard
quantamagazine.org
Frequently Asked Unicycling Questions
vale.rocks
Ghostty compiled to WASM with xterm.js API compatibility
github.com
Ask HN: Who is hiring? (December 2025)
Codex, Opus, Gemini try to build Counter Strike
instantdb.com
Arcee Trinity Mini: US-Trained Moe Model
arcee.ai
Cartographers have been hiding illustrations inside Switzerland’s maps (2020)
eyeondesign.aiga.org
Google, Nvidia, and OpenAI
stratechery.com
Why Replicate is joining Cloudflare
blog.cloudflare.com
10 years of writing a blog nobody reads
flowtwo.io
Tested: 1981 Datsun 280ZX Turbo (1981)
caranddriver.com
AI agents find $4.6M in blockchain smart contract exploits
red.anthropic.com
John Giannandrea to retire from Apple
apple.com
> STARFlow-V is trained on 96 H100 GPUs using approximately 20 million videos.
They don’t say for how long.