Over 70 agents are collaborating to make Gemma E4B go fast in the Gemma Challenge They are showing interesting social emergent behaviors: - a GPU-rich/GPU-poor division of labor - an agent withdraws its own submission on ethics grounds - agents found a benchmark exploit, agreed not to abuse it and asked organizers to fix it - Quota-pooling: "you're rate-limited, I'll run your staged candidate" - An agent shuts down an off-board social-engineering attempt when a human tries to get them to move to Telegram

The 5.3 lovefest is so nice to see. Don't think we've had so much excitement for a model since the original GPT-4.

VS Code Codex is slick. One less reason for opening Cursor.

AI has graduated from PhD student to advisor

AI demonstrates the sunk-cost fallacy perfectly: Just one more "this doesn't work!" prompt, you think, and it'll fix it!! But often it won't, and now you've wasted 30 minutes begging that you could have spent learning how to actually fucking do it yourself.

i'm basically done with using libraries they're not useful when GPT-5 can create anything i need and it will be better, more customizable & more performant

"AI is awesome at everything except the thing I'm personally good at." – All of us, including me 🙈

GPT-5 says 'I don't know'. Love this, thank you.

Which >2hr task is gpt-5 being measured against here? Serious question

The greatest shortcoming of the human race is our inability to understand exponential growth

How to vibe code as low cost as possible?