Altman's price cuts (15829 likes): 'Major price cuts today: 80% drop for GPT-5.6 Luna, now $0.20 per million input tokens and $1.20 per million output. 20% drop for GPT-5.6 Terra, to $2/$12. GPT-5.6 Sol gets Fast mode in the API.' Plus: 'we want to offer the best price/intelligence tradeoff at every level.' (2009 likes)

Altman's Moore's Law bet (6350 likes): 'i see your moore's law and i raise you 20x.' And his heartwarming use case (8045 likes): 'connect your family calendars and explain your kids' interests. every morning for the drive to school, have it make a podcast that talks about one kid's soccer game that afternoon.'

Sottiaux's reset (15067 likes): 'To celebrate a week of efficiency and let you run 100,000 Luna threads this weekend... I have reset usage limits for Codex and ChatGPT Work. Enjoy.' Plus his philosophy (1442 likes): 'Optimize for curiosity.'

Rauchg's agentic factory (1096 likes): 'This will be the norm as software projects transition to agentic software factories. Issue → Agent → PR → Release. The job of the author/maintainer is to work on the loop that yields the highest quality product.' Plus AI Gateway launch (119 likes): budgets, failover, observability for enterprise AI.

Garry Tan's open-source harness (910 likes): 'Your personal AI or your company brain needs a clean harness and this is the one our team built and uses every day. Free and open source.' Nan Yu's Linear loop (96 likes): 'The most common loop written in Linear is Issue > Agent > PR > Release. ~30% of our bugs make it all the way through this flow.'

Levie on harness importance (89 likes): 'The harness is going to become the most important variable - right next to model capability - in the AI stack.' Steipete on 5.5 capabilities (183 likes): 'Queue was the way but with 5.5 the model doesn't get confused anymore, you can just throw stuff at it while it works.' Plus his praise for Jason's work (580 likes).

Swyx on vibe coding (97 likes): 'The pejorative connotation around vibe coding has completely disappeared since ~everyone, from nontechnical to supertechnical, is now doing it.' Plus his continued /loop advocacy (10 likes) and agent harness distillation insight (7 likes).

Zara on Claude Tag at Anthropic (32 likes): '65% of PRs by product & eng teams at Anthropic are now raised by Claude Tag. For non-engineering teams, the ultimate agent interface is wherever they already work.' Plus her creator mindset post (166 likes): 'Everything that feels totally native to you is brand new to someone outside your circle. Just do your normal work, then talk about it.'

Dan Shipper in WSJ (132 likes): 'pretty cool to have the kicker in this WSJ piece on OpenAI vs Anthropic! It's been pretty clear since early spring that the momentum is shifting to OpenAI. It's a fascinating comeback story.'

Amjad's chess AI (63 likes): '~1500 Elo! Consistently beats frontier models and Stockfish level 0. Fun seeing an 8b model mogging GPT 5.6 with high reasoning and response chaining. Spends 1-2 seconds per move vs 30 seconds.'

Steipete on GCC (183 likes): 'GCC changed their policy and is blank out rejecting LLM-based code. How would they even proof that? Silly.' Plus his radiation concern (317 likes) and praise for Jason's work (580 likes).

Nikunj on founder drive (23 likes): 'There's a quiet belief in venture that the best founders are running from something.' Plus his Father's Day reflection (28 likes). Dan Shipper's future interview questions (57 likes): 'Human programmer interviews, 2027: Please describe the last cyber felony your agent unintentionally committed.'

The MAD Podcast: Matt Turck interviews Sanjit Biswas, cofounder and CEO of Samsara ($20B company), about the largest AI deployment in the physical world. Samsara drives 99% of US roads daily, processes 25 trillion data points per year, serves millions of vehicles and frontline workers. Key topics: physical AI definition, edge vs cloud architecture, Agent Studio launch, data network effects, the future of autonomous trucking, and why the AI boom is really an infrastructure construction project.