closing the loop is key for exponential improvement
nice work!
Amjad Masad: Many are asking how Replit is improving so rapidly—we closed the loop and the agent is self-improving.
Technical details here:

Hermes, OpenClaw, Claude Code doesn't need more Memory. They need TASTE.
Here's a step-by-step guide to build that with @garrytan's Gbrain.
Shubham Saboo: http://x.com/i/article/2074343494005764098

Re @davidsenra @JonathanRoss321 Pretty sure this was the post:
Matt Shumer: Wild tech you have to try: https://groq.com
They are serving Mixtral at nearly 500 tok/s.
Answers are pretty much instantaneous.
Opens up new use-cases, and completely changes the UX possibilities of existing ones.

the booth was a great investment
Theo - t3.gg: I would like to add that OpenAI has been awesome to work with despite the obvious government-related setbacks.
They've been more transparent than I'd ever have expected, and they did a good job keeping us in the loop during the confusion.
The only "restriction" is that they've

Re @atgorans_k Okay ig we can share now. This was also my experience. Full throttle.
https://x.com/mattshumer_/status/2074720356510609788?s=20
Matt Shumer: I had early access to GPT-5.6 Sol.
It’s an amazing model, but for almost every task I tested, Fable was quite a bit better, and more agentic to boot (one Fable turn does the same thing many 5.6 turns do).
Full review coming on launch day!

I thought I'd be done processing all of @RoKhanna's financial disclosures tonight but the agents are STILL grinding through his tens of thousands of assets.
Three years are uploaded at https://rokhanna.money
Kane 謝凱堯: Rep @RoKhanna has been playing populist w performative policies like banning Waymo.
Instead of Congress's electronic financial disclosures, he hand-files papers that are not searchable.
I burned some tokens and OCR's all >1000 pages and >$300M of his finances! Site shortly.

My big takeaway is that both Sol & Fable represent jumps over previous models and have opened a large gap with the next-best AIs. People will have preferences for one or the other, but if you doing any work where better intelligence matters, those two models are your only choices

For the past year, MAGA has been screaming “Europoor” like a cheap car alarm going off at three in the morning. Nobody asked for it, nobody knows how to turn it off.
But then along comes this chart.
America sits at number four in average wealth. Except “average” is the statistic billionaires hide behind. Put Elon Musk in a room with 500 broke MAGA guys, and on average, everyone in that room is a multibillionaire. That’s the trick!
Now look at the median. The actual middle American, the real one with the pickup truck, the missing teeth and the medical debt. Fourteenth place. Behind Belgium. Behind Italy. Behind Spain. And the ordinary European? The man MAGA calls a Europoor pays $9 for his insulin, glances over at the American paying $900 for the exact same vial, shrugs, and goes back to being more than twice as wealthy. He’d feel sorry for you, but honestly, he’s too busy being on vacation.
And that gap between fourth and fourteenth? That’s not a statistical quirk. That’s the entire scam laid bare. America’s wealth exists. It’s just not yours. It belongs to about eleven people in Palo Alto, and you’ve been guarding it for them for free, like an unpaid nightclub bouncer, shouting at Europeans on the internet while the owners drink champagne inside.
American exceptionalism. No other nation on Earth has ever been this successfully convinced to cheer for its own fleecing.
So by all means, keep shouting Europoor. We’ll be over here with our healthcare, our six weeks of vacation, and our double-your-median wealth, trying very hard to hear you over the sound of how sorry we feel.

http://x.com/i/article/2074768591900098560

We're releasing ZML/LLMD, our homegrown LLM server built on top of our homegrown high performance heterogeneous inference stack.
It ships with 5 architectures out of the box: NVIDIA, AMD, Metal, Intel and TPU. All transparent.
It supports DFlash, continuous batching, prefix caching, the whole deal.
Oh, and it's fast.
ZML: http://x.com/i/article/2074768591900098560

The test of a description of a product is how much closer I am after hearing it to being able to reproduce it. So e.g. "transform the way people interact with images" has almost zero descriptive value. If I had to make this, where would I even start?

Who else will be happy when San Francisco Supervisor Shamann Walton is termed out in a few months and can go back to his actual home in Vallejo? Vote for @theoellingtonSF to replace this corrupt clown. https://www.sfchronicle.com/sf/article/supe-accepted-painting-agrees-to-fine-22283111.php?taid=6a4c638657b3960001506f01&utm_campaign=trueanthem%2B3988&utm_medium=social&utm_source=twitter

We are excited to launch 𝗥𝗲𝘀𝘁𝗮𝘁𝗲 𝗕𝗬𝗢𝗖 (Bring-your-own-Cloud) today.
It is fully managed @restatedev, with all the features of Restate Cloud, but in your own account, in a dedicated VPC. Data never leaves your infra.
Restate BYOC is already live in production with multiple customers, powering big workloads for AI agents and workflow backends. It has run loads of 100k durable actions/sec. And today we’re finally publicly announcing it.
Learn more here, and let us know what you think!
https://www.restate.dev/blog/announcing-restate-byoc?utm_source=x&utm_campaign=byoc-launch&utm_content=founder

New this week 🚀 Replit Community Profiles — Proof of Work for vibe coders.
Your profile, your flex. Get an activity graph of your agent usage and checkpoints, plus a Replit Power Ranking for pro users.
Log in, claim your profile, pick your best projects, and share your stats with friends.
See it in action now → http://replit.com/community

Sol for developing next.js:
Tim: We've been testing GPT-5.6-Sol for over 2 months now. It’s incredibly good in my day-to-day working on Next.js.
It understands architecture tradeoffs. It can investigate complicated Next.js issue reports. It considers other areas of the codebase when fixing bugs. Needs very

Introducing Tavus Google Meet integration.
Now a PAL can join a Google Meet, a real participant in the call you were already going to have, taking its turn the way any other guest would.
Make and deploy a customized notetaker, personal assistant, or even a meeting host in a few clicks.

A free custom domain, on us.
Build, publish, then give your app a name, at no extra cost until July 17.

We raised $130M @ $1B for our series A
To build the open superintelligence stack for everyone
Pre-training concentrated frontier AI in a handful of labs. RL changes who can build frontier AI and just works across almost any verifiable domain. We want to enable everyone to train their own agents.
Companies can now own their model optimization loop: train directly on your product, optimize for your specific workflows, and build agents that improve continuously in production
Owning this model <> product improvement loop is how you build a compounding moat in the agentic era
Super grateful to serve over 6k+ customers, including many leading AI startups, neolabs and enterprises already building on our stack, and to our incredible team for shipping hardcore!
We train open frontier models and ship the same stack to our customers. Its spans the full stack of training, deploying and continuously improving models — compute, large-scale RL, environments, sandboxes, evals, and deployment.
We're excited to be joined by angels who are building the frontier themselves, many of whom we work closely with:
@johnschulman2 (Thinking Machines), @dwarkesh_sp, @AravSrinivas (Perplexity), @karimatiyeh (Ramp), @levie (Box), @_milankovac_ (Tesla), @winstonweinberg (Harvey), @amspector100 (Flapping Airplanes), @jeffwang (Cognition), @_arohan_ (Core Automation), @marksaroufim (Core Automation), @mikeknoop (Zapier, Ndea), @eastdakota (Cloudflare), @BrendanFoody (Mercor), @devanshpandey (Standard Intelligence), @hwchase17 (Langchain), @nicoup (Fleet) and many more
We're a small team building open superintelligence
> Reach out if you want to partner training, deploying and continuously improving your own frontier models for your use case
> Join us to build open superintelligence — we're hiring across all roles including RL, inference, distributed systems, full stack engineering and compute.
Prime Intellect: Announcing our $130M Series A to build the Open Superintelligence Stack
Led by Radical Ventures, with NVIDIA, Intel Capital, Dell Capital, and existing investors
Train, deploy, and continuously improve your own models using our stack.
Own your intelligence.

Someone asked me what Screen Recorder I use.
I use Screen Charm but I've just built my own one in @Replit for $2.02 that records system audio and you can also select the screen area, rather than the full window.
I used Screen Charm to record my Replit Screen Recorder. 😂
#Replit #AdeviousAI #BuildInPublic #ScreenRecorder

Today we're launching Glass for Patients, the personal AI agent for your health.
Glass reviews all of your healthcare data and analyzes it alongside the world's medical knowledge to help you continually understand and optimize your health.
Get started at http://glass.health/for-patients

We're still just scratching the surface of how much better, faster, cheaper we can make inference for open-source models!
So super excited to see companies like @zml_ai by @steeve come out of stealth with their inference engine integrated with @huggingface as the storage layer.
Let's push open-source tokens to the max!

http://x.com/i/article/2074573342187868160

We've usually stayed away from model comparisons but 5.6 vs Fable is a unique situation
We've never had a case where the team is so completely convinced on which one is better
Here's the timeline of our experience with it
- We test early versions of 5.6 for a couple of weeks and have a great time, it feels like a step change improvement, enabling new workflows
- We get to try Fable and don't think it's not as good, I personally would take this experience with a grain of salt, there tends to be a bias when trying a new model when you already like another
- Fable and 5.6 are taken away because of the regulatory issues
- Our team is literally depressed that 5.6 is gone, we are looking for anything that could even partly replace it
- Fable comes back, and here's where it gets interesting, you would think Fable would be enough, but no, the team is still depressed that 5.6 isn't available
- Then 5.6 comes back and it's immediately clear that it's just way better than Fable
This situation was unique in that it was the closest we've ever gotten to having an unbiased comparison of two models

The campaign against American AI runs directly from Shanghai to San Francisco.
California is both the home of US AI development & a staging ground for a CCP-aligned campaign to snuff it out.
America's adversaries are boosting the degrowth movement. A must-read from @garrytan
Garry's List: A Shanghai-based millionaire funneled $285M into US nonprofits now under federal investigation for fraud and money laundering. His wife spent February in Oakland organizing to kill American AI data centers.
We used research from @SamLyman33 and @bitcoinpolicy to show how

The emoluments clause is supposed to prevent the President from receiving foreign bribes.
That hasn’t stopped Trump’s family from receiving about $13 billion in “investments” from foreign governments, especially in the Middle East.
My @Morning_Joe Chart

Try out Grok 4.5 in Vercel!
Guillermo Rauch: Grok 4.5 now available to all Vercel customers

San Francisco and San Jose pivoted their homeless policies toward accountability and recovery. They're beginning to see results. Check out my op/ed in @CityJournal here: https://www.city-journal.org/article/san-francisco-san-jose-homeless-encampments

I had a pretty similar experience
A bunch of my tweets about missing Fable were actually about missing GPT 5.6 Sol, but I couldn’t say it so said Fable instead
Fable is really good and maybe better in some ways, but I vastly prefer GPT 5.6 Sol
Jay: We've usually stayed away from model comparisons but 5.6 vs Fable is a unique situation
We've never had a case where the team is so completely convinced on which one is better
Here's the timeline of our experience with it
- We test early versions of 5.6 for a couple of weeks

Under Trump, pardons have a price, and criminals know it.
Trump has erased a growing number of criminal convictions – and $1.5 billion in penalties and restitution owed to taxpayers and victims – for his donors and friends.
My @Morning_Joe Chart

Rewriting Bun in Rust
https://bun.com/blog/bun-in-rust

Just 12 years ago, the New York Times wrote “Oakland: Brooklyn by the Bay”— a feature story celebrating that Oakland might turn into a top flight city. Sadly, after a succession of failed progressive policies, mayors, city councils, and city hall corruption scandals, Oakland is no longer on the verge being like Brooklyn. It’s more like “Escape from New York.”