Product Moves (Big-Lab Competition) 3
Anthropic: Fable 5 extended + Claude Code limits +50%
Clear competitive response to GPT-5.6; community both thankful and joking "Fable 5 extended for the Nth time"
OpenAI: Codex 5-hour cap removed, ChatGPT Work updates
Letting GPT-5.6 Sol + Codex run "unlimited" to win developers
GPT-5.6 Sol reception
Broadly positive; sama runs a "build the coolest thing with 5.6 Sol, win a prize" call
Twitter / KOL Discussion 4
Sam Altman: AI is net job-creating so far
Against his prior expectation; sparked a big "is it really?" debate
geohot: I love LLMs, I hate hype
George Hotz separates "LLMs are genuinely useful" from the hype bubble; HN 333 pts, 200 comments
pmarca: AI is already a better doctor than 99.99% of humans
Marc Andreessen's provocative claim
emollick: Google Maps quietly A/B-tests routing
Google routed 2% of drivers via "just as fast but greener" paths; Mollick on the ethics of silent large-scale experiments
Tools & Discussion (HN) 5
Claude Code sends 33k tokens before reading your prompt
Measured overhead vs OpenCode; #1-ish today at HN 474 pts, 266 comments
Terence Tao: old and new apps via modern coding agents
Fields medalist blogs his hands-on with AI coding agents; HN 417 pts
Migrating a production agent to GPT-5.6: 2.2x faster, 27% cheaper
Real-world migration data
Agent Harness Engineering
Addy Osmani on building reliable agent scaffolding (tools, context, flow)
Mindwalk — replay coding-agent sessions on a 3D map
Show HN visualizing agent steps
HF Trending Papers 3
LongE2V
Long-horizon event-based video reconstruction & prediction (28↑)
DrugGen 2
Disease-aware language model for drug discovery (17↑)
Proactive Memory Agent
Remember only when it matters, for long-horizon tasks (10↑)
Podcasts 1
[No Priors]
No Prior Warning: Atlas Ends, MuseSpark Arrives (Jul 10)