<< Go back to HN Digests
Warning : This document is still a draft

2026 W16 - 04-19 - Opus 4.7




First time here ? Check the introduction here

What you Missed this Week

Opus 4.7 release ! It cost on average 30% more tokens than 4.6 … (Link)

Articles

Model Training / Tunning

Model improvement or other low level optimization

  • 1955 pt [Link / HN] Claude Opus 4.7
  • 712 pt [Link / HN] Measuring Claude 4.7’s tokenizer costs

  • 176 pt [Link / HN] Claude Opus 4.7 Model Card

  • 280 pt [Link / HN] Introspective Diffusion Language Models
  • 216 pt [Link / HN] Ternary Bonsai: Top Intelligence at 1.58 Bits

Coding Agents

Claude new functionalities, tips and tricks dedicated to a specific agent, …

  • 1230 pt [Link / HN] Claude Design

  • 718 pt [Link / HN] Claude Code Routines

  • 339 pt [Link / HN] ChatGPT for Excel

Personal Blogs / Opinions

Investigation

Trend investigation, hacking, data analysis, …

  • 198 pt [Link / HN] We gave an AI a 3 year retail lease and asked it to make a profit

Tool for AI / Made with AI / showHN

Any Github / ShowHN for new tools / product

Optimists

  • 642 pt [Link / HN] The local LLM ecosystem doesn’t need Ollama
  • 462 pt [Link / HN] Qwen3.6-35B-A3B on my laptop drew me a better pelican than Claude Opus 4.7
  • 437 pt [Link / HN] Apple’s accidental moat: How the “AI Loser” may end up winning
  • 365 pt [Link / HN] Changes in the system prompt between Claude Opus 4.6 and 4.7

Pessimists

Other

  • 383 pt [Link / HN] Thoughts and feelings around Claude Design

  • 362 pt [Link / HN] I’m spending months coding the old way
  • 239 pt [Link / HN] AI cybersecurity is not proof of work

Officials

Politics / Strategy

Merge, acquisition

News

Anything from “official” journals

Unrelated

Not related to AI, but worth reading

  • 990 pt [Link / HN] Stop Flock
  • 739 pt [Link / HN] The future of everything is lies, I guess: Where do we go from here?
  • 613 pt [Link / HN] Anonymous request-token comparisons from Opus 4.6 and Opus 4.7
  • 558 pt [Link / HN] Cybersecurity looks like proof of work now
  • 500 pt [Link / HN] Darkbloom – Private inference on idle Macs
  • 461 pt [Link / HN] Cloudflare Email Service

  • 417 pt [Link / HN] The economics of software teams: Why most engineering orgs are flying blind

  • 404 pt [Link / HN] The “Passive Income” trap ate a generation of entrepreneurs
  • 379 pt [Link / HN] Vercel says internal systems hit in breach
  • 351 pt [Link / HN] The RAM shortage could last years
  • 345 pt [Link / HN] This year’s insane timeline of hacks
  • 344 pt [Link / HN] Everything we like is a psyop?
  • 336 pt [Link / HN] Building a CLI for all of Cloudflare

  • 306 pt [Link / HN] Cloudflare’s AI Platform: an inference layer designed for agents

  • 330 pt [Link / HN] The Future of Everything Is Lies, I Guess: Safety
  • 289 pt [Link / HN] The future of everything is lies, I guess: Work

  • 274 pt [Link / HN] The Future of Everything Is Lies, I Guess: New Jobs

  • 314 pt [Link / HN] Android CLI: Build Android apps 3x faster using any agent
  • 305 pt [Link / HN] Are the costs of AI agents also rising exponentially? (2025)
  • 272 pt [Link / HN] Sometimes powerful people just do dumb shit
  • 263 pt [Link / HN] Stanford report highlights growing disconnect between AI insiders and everyone

  • 259 pt [Link / HN] Codex Hacked a Samsung TV

  • 253 pt [Link / HN] Does Gas Town ‘steal’ usage from users’ LLM credits to improve itself?

  • 230 pt [Link / HN] AI-assisted cognition endangers human development?

  • 213 pt [Link / HN] An AI Vibe Coding Horror Story
  • 210 pt [Link / HN] Laravel raised money and now injects ads directly into your agent
  • 197 pt [Link / HN] US v. Heppner (S.D.N.Y. 2026) no attorney-client privilege for AI chats [pdf]
  • 195 pt [Link / HN] The beginning of scarcity in AI
  • 189 pt [Link / HN] Show HN: I built a social media management tool in 3 weeks with Claude and Codex
  • 186 pt [Link / HN] Dependency cooldowns turn you into a free-rider
  • 186 pt [Link / HN] AI could be the end of the digital wave, not the next big thing
  • 176 pt [Link / HN] Show HN: Ithihāsas – a character explorer for Hindu epics, built in a few hours
  • 175 pt [Link / HN] I learned Unity the wrong way
  • 172 pt [Link / HN] Dad brains: How fatherhood rewires the male mind
  • 171 pt [Link / HN] Kefir C17/C23 Compiler
  • 165 pt [Link / HN] Zig 0.16.0 Release Notes
  • 159 pt [Link / HN] Show HN: MacMind – A transformer neural network in HyperCard on a 1989 Macintosh
  • 156 pt [Link / HN] GAIA – Open-source framework for building AI agents that run on local hardware
  • 156 pt [Link / HN] Show HN: Prompt-to-Excalidraw demo with Gemma 4 E2B in the browser (3.1GB)
  • 148 pt [Link / HN] Show HN: LangAlpha – what if Claude Code was built for Wall Street?
  • 148 pt [Link / HN] Show HN: MDV – a Markdown superset for docs, dashboards, and slides with data
  • 134 pt [Link / HN] Claude may require identity verification in some cases
  • 132 pt [Link / HN] SDL bans AI-written commits
  • 123 pt [Link / HN] State of Homelab 2026
  • 120 pt [Link / HN] Show HN: SPICE simulation → oscilloscope → verification with Claude Code
  • 120 pt [Link / HN] Silicon Valley is turning scientists into exploited gig workers?
  • 119 pt [Link / HN] Zero-Copy GPU Inference from WebAssembly on Apple Silicon
  • 119 pt [Link / HN] Multi-Agentic Software Development Is a Distributed Systems Problem
  • 119 pt [Link / HN] The AI revolution in math has arrived
  • 118 pt [Link / HN] OpenAI’s $852B valuation faces investor scrutiny amid strategy shift, FT reports
  • 118 pt [Link / HN] When moving fast, talking is the first thing to break
  • 113 pt [Link / HN] Graphs that explain the state of AI in 2026
  • 113 pt [Link / HN] Average is all you need
  • 112 pt [Link / HN] Scan your website to see how ready it is for AI agents
  • 112 pt [Link / HN] Gas Town: From Clown Show to v1.0
  • 110 pt [Link / HN] Show HN: CodeBurn – Analyze Claude Code token usage by task
  • 109 pt [Link / HN] We reproduced Anthropic’s Mythos findings with public models
  • 107 pt [Link / HN] Can Claude Fly a Plane?
  • 106 pt [Link / HN] Prove you are a robot: CAPTCHAs for agents
  • 105 pt [Link / HN] Allbirds announces pivot from shoes to AI, stock explodes 175%
  • 102 pt [Link / HN] GPT‑Rosalind for life sciences research
  • 101 pt [Link / HN] Banned by Anthropic?
  • 100 pt [Link / HN] CPUs Aren’t Dead. Gemma2B Out Scored GPT-3.5 Turbo on Test That Made It Famous
  • 96 pt [Link / HN] Show HN: Plain – The full-stack Python framework designed for humans and agents
  • 94 pt [Link / HN] Uber’s Anthropic AI push hits a wall
  • 94 pt [Link / HN] Two Months After I Gave an AI $100 and No Instructions
  • 86 pt [Link / HN] TanStack Start Now Support React Server Components
  • 82 pt [Link / HN] George Orwell Predicted the Rise of “AI Slop” in Nineteen Eighty-Four
  • 82 pt [Link / HN] The human cost of 10x: How AI is physically breaking senior engineers
  • 82 pt [Link / HN] Programming Used to Be Free

>> You can subscribe to my mailing list here for a monthly update. <<