3 crazy things we learned from ViBench about current models
1/ Frontier models are now surprisingly close at the top, but there is still one lead π
2/ There's still a massive gap between the best and the rest. π€
3/ Price and performance aren't perfectly correlated. π
The best metric is price/performance, which we highlight in our benchmark!
Benchmarks place GPT 5.5 as the best model on SWE, but is it the best at making apps end-to-end?
Turns out Opus 4.8 continues to be the king of vibe coding on both price & performance.
Introducing ViBench: the first benchmark for app creation based on real world tasks
It was a pleasure to collaborate with the team at @Microsoft on this.
This is not the Microsoft that you remember.
It was mindblowing to me how ambitious and startup-minded their engineering team is.
Truly humbling!
Excited to partner with @Microsoft to enable everyone in the enterprise to build and deploy safe & secure Fabric data apps.
This is possible thanks to Microsoft's new Rayfin SDK.
I will soon be introducing a bill to give the public a 50% ownership stake in the largest AI companies in America.
This would guarantee that the trillions created by AI are used to improve the lives of all of us β and block oligarch decisions that harm the American people.
Can you build a real business for free with a single prompt?
Starting today on Replit, the answer is yes.
From a single prompt, get a website, mobile app, slide deck, and launch video.
Plus unlock perks from @stripe@atlas, @QuickBooks, @mercury & @doolaHQ
Sometimes in life, motivation dips, even when the path still matters to you.
Donβt let the first or second slump decide the outcome. A low day, even a heavy one, doesnβt mean the quest is wrong.
It just means you keep going gently, step by step, until the spark catches again.
βKids will never be smarter than AIβ is the wrong frame.
Kids born today grow up with a tutor, builder, translator and research assistant in the loop.
Not dumber. Just starting with better tools than the 2000s ever had.
https://t.co/2WliZSDL0Z
The best design work doesn't happen in a chat box. You need space to explore ideas, create variants, and iterate
Meet the new Replit Canvas
Your agentic design tool to build beautiful websites, apps, marketing assets and more
14.5M views on Altman describing intelligence as electricity or water.
The trap is treating the meter as the business.
The valuable stuff is what builders plug into it.
https://t.co/htHGBHTVxP
Michele Catasta (@pirroh) is President and Head of AI @replit, the platform where anyone can build software in natural language.
At 16, he set out to make software open to everyone. Today, over 50 million people are building on Replit with Claude:
YC is quietly building a counter-drone ecosystem.
@PerseusDefense - Guided missiles to shoot down drones
@9Mothers - AI machine guns to shoot down drones
@surtrdefense - Open OS that fuses every sensor into one threat picture
Milliray - Spots drones too small for legacy systems
Arlo Industries - Passive aerial sensing mesh for drone detection
Three years ago, I couldn't get the best defense founders to apply. Now I can barely keep up πΊπΈπͺπΊπΈ
@arnavsahu341 I have some background inference classification tasks on their PRO model yes. I don't use it as my daily coding driver though. Super cheap and fast.