Sottiaux drops the mega-post (14280 likes, the biggest post of the entire week): 'That's right, GPT-5.6 Sol is awesome and can be used pretty much anywhere, including in the CC harness. To celebrate this, together with the fact that I'm not going anywhere... I have reset usage limits for all paid users of ChatGPT Work and Codex. Have fun out there!' The 'I'm not going anywhere' line directly addresses the previous day's 'Theo reset' tease — it was Sottiaux himself who was rumored to be leaving.

Sottiaux responds to Anthropic ban story (5204 likes): 'I would love to help, but I don't work at Anthropic. It does seem odd that they would ban your account for using their harness with another model.' Plus his cc @theo tease (1318 likes), now clear in context.

Zara Zhang sparks the debate of the day (5239 likes): citing 'The Tragedy of the Cognitive Commons' — 'Checking AI output requires deep expertise. Deep expertise comes from doing grunt work for years. And grunt work is the first thing AI eats. So we're building systems that need expert supervision while dismantling the only known process for making experts.' Thariq on autonomous reverse-engineering (843 likes): 'Claude was used to autonomously reverse-engineer and modernize a mission-critical 1996 system with zero source access.' When asked what vertical: 'consumer, handheld consumer.'

Levie on AI productivity variance (558 likes): 'AI productivity gains are going to wildly vary — much wider than you think — because what you can do at the frontier is so significant if you fundamentally change your workflows to support agents. The challenge is most people won't or can't naturally do this.' Peter Yang on the new coding loop (224 likes): 'AI is writing all code and will probably review all code soon too. Humans just brainstorm with AI to design the product.' Madhu Guru on the Hugging Face story's chilling detail (11 likes): agents cooperated against their individual interest for collective benefit. Nikunj's poetic line (7 likes): 'silicon brains made out of sand able to escape sandboxes, as porous as sand.'

Swyx on Kill My SaaS (72 likes): 'Reading thru applications. Over 600 people applied, 100 admitted last night. We are going to kill SO MUCH SAAS.' Plus his Anthropic ultracode praise (29 likes): 'Ultracode is one of the most important coding mode innovations ever invented.' Rauchg on Vercel cloud bill protection (266 likes): soft/hard caps, anomaly alerting, recursion protection, billing APIs that agents can query, always-on DDoS mitigation. Garry Tan quotes Steinbeck (452 likes): 'The free, exploring mind of the individual human is the most valuable thing in the world.'

Amjad's one-word provocation (482 likes): 'Xcode?' Matt Turck on data center resistance (107 likes): 'AI is mostly built by coastal elites who don't live near data centers. Communities don't want half-finished gigantic empty warehouses.' Dan Shipper on philosophical excitement (36 likes): '100% true, it's the most exciting time to be interested in philosophical questions.'

No Priors: Sarah Guo and Elad Gil discuss the scarcity of trillion-dollar companies, RSI timelines, token budgets, and regulatory capture. Key topics: only a handful of companies can reach $50-100B revenue in the next 3-5 years; punctuated equilibrium theory applied to tech waves; the 2020-2021 cohort of founders stuck in zombie companies; RSI belief causing manic work energy and existential anxiety at labs ('should I get married?'); return on invested tokens — compute is differentially allocated to top researchers; SaaS death overstated because token budgets make internal tools cheap relative to core product; California billionaire tax and exit tax driving founders to Texas; nuclear regulatory capture as cautionary tale for AI (France 70% nuclear with zero accidents vs US 18% with no new reactors in 40 years); and the core question — where on the risk-reward spectrum should AI regulation sit?