Suhail Doshi

Steadily growing every day! More and more people keep finding particle-1 great to use due to token efficiency.

Particle has so many similarities to Mixpanel (scaling events vs tokens). Sometimes it’s like de ja vu. I’ll often feel like I had already gone through a problem before. Since I’ve had a lot of space away from the first co, things feel fresh again.

Woke 2.0 has begun.

We have doubled our compute capacity as of today. 🚀

See? If you just waited 2 weeks before taking action as a board, everything can be different.

Once again this is why there should be rule about no action on the CEO for 2 weeks (unless criminal) by the board during a PR crisis.

RT alon Introducing Current, an opinionated way to work with your agents. Current lets you run hundreds of agents while staying in control. I've built everything with it over the past couple of months, including Current itself.

Jonathan Ross who led Groq / TPUs. He’s is a 🐐

There should be a rule on boards that you do not make any moves on CEO for 2 weeks after a PR crisis (excluding criminal acts). Boards are often compelled to make decisions under public pressure but once that subsides there's a great clarity about what to do.

Nobody should be surprised by this. This is why I started Particle—there must be an answer to taxing humanity on intelligence. We will drive the cost per task to zero.

GTABench: what % of GTA can the model re-create? Start with GTA 1 and hill climb to GTA 6. 100% accuracy fully re-creates the game that would be indistinguishable from the original version for players.

The last pre-agi cultural event of our species. 🫡

Our model particle-1: "A review costs us 2 to 3 cents." We're going drive cost per task down to zero!

We'd like to partner with a Agentic Cowork-style tool (docs, slides, pdfs) with low-cost AI inference. If you're a startup, please reach out. We will only consider people with @ min 10K WAU. DM.

I think the folly of Anthropic is believing you can control AI at all.

The capitalist in me gets it but this won’t go unanswered by society. Anthropic will continue to gatekeep until your products can’t use their models or you can’t make certain things that they’ll compete with you on.

Thank you @bchesky City of the future!

🎯

There aren’t really any data advantages in frontier AI imo. There’s really just capital advantages. But once the data budget exceeds $10B/year, it’s back to taste and people.

I’ve always wanted to do this test but never can get around to it. Glad someone did it!

yep

Very important graph at *log scale* credit: @natolambert

RT sina Just tried particle-1. @Suhail is cooking with gas.

I wonder if we will find it interesting as humans to watch AIs compete with one another in various ways.

🚀🚀🚀 Next model is baking in the oven.

We'd like to partner with a Agentic AI coding tool with low-cost AI inference. If you're a startup, please reach out. We will only consider people with @ 10K DAU. DM.

If you are working on tools and agents to do security agents, I would love to talk to you. We have something very cool cooking. Pls dm.

It is now table stakes to be running a 24/7 process of scanning for vulnerabilities and attempting to self-pwn your software/infra. This will be a compliance requirement for any business doing business with others.

🎯

RT Garry Tan Was a true honor to sit down with @WalterIsaacson for The Builders podcast out this morning We are at an unusual inflection point in history and there are so many reasons to be excited for the future with superintelligence almost here

Now in OpenClaw - it was collaborating with them on this! We have a lot more updates to the model coming...

RT Spenser Skates Meeting with my archnemesis @Suhail We spent 10 years battling it out in analytics between Amplitude and Mixpanel. Surprisingly, we have much more in common than expected. I’m learning marketing from him and he’s learning sales from me.

Pareto frontier on data analytics. Honestly feels like a full circle moment. 🫡

Interviews for this begin this week. I think this is a really fun role for that blurs the lines between applied ai research and product engineering. Great for someone who likes focusing on building for developers.

We are going to drive cost per task to zero. 🫡

RT Garry Tan Particle-1 is as good as much larger models for agentic retrieval but 1/23 the price https://github.com/garrytan/gbrain-evals/blob/b392700f2f647c201a8bedca71d43fd6a7fbb31c/docs/findings/2026-11-01-particle-1-longmemeval-w10b.md

RT Kyle Anthony Miller Big footer energy

Congrats to GSV! ✊

“Once man has created an intelligence greater than his own, that intelligence should be able to create an intelligence greater than itself, as man whose intelligence is less than its was able to create its intelligence.”

Do you have access to a cyber AI model (to defend yourself) by one of the closed frontier labs?

Easily the coolest job for a hardware person.

Excited to be partnering with Cloudflare and on their AI gateway - try our new token efficient model. 🚀

🎯

RT Garry Tan I've been using particle-1 and very impressed

Independently validated in real-swe!

RT Moin Nadeem Introducing Phonic-1, the most advanced speech-to-speech model for agentic tasks — from @phonic_co, the speech-to-speech model company. We’ve had a breakout year and raised a $27.5M Series A led by @firstround with participation from @Lux_Capital, @felicis, and Definition.

We are going to drive cost per task to zero. We will not allow a duopoly to tax humanity on intelligence. We trained on humanity’s data, the least we could do is give back and let everyone prosper.

I am very excited to see someone independently validate our claims on something we couldn’t train on and is real-world. Same performance capability, with way less tokens.

RT Tal Hoffman We ran particle-1 through our Hacking Race today. While it was trained for coding and data work, not security, it still got working exploits on 2 of our 4 targets, with zero false positives on the patched versions. Strong first model from @Suhail and the http://particle.ai team

RT Aiden Bai pretty impressive results on ReactBench!

Thank you everyone for your support! 🚀🫡❤️

Very excited to announce partnering with @OpenRouter and launching our Inference service. particle-1 is live! Please send us feedback. Will respond to all DMs if you find any issues. Happy to fix things.

Boom 🚀 Excited to partner with Vercel and @rauchg

RT Sidharth Raja one thing i can absolutely say for certain after signing up to try, is that this was the fastest "time to api key" experience i've had for anything i've ever used

"Now all I get is hate mail all day sayin' Dre fell off What, 'cause I been in the lab with a pen and a pad Tryna get this damn label off? I ain't havin' that This is the millennium of..."

Tomorrow.

New quest. New company. New model. It's time to ship this week.

Moravec's paradox afflicts much of general robotics but I am quite optimistic overall—it may take more time than it seems.

It's time to ship.

🎯 Inference is not the same shape as a neocloud. You have many more levers to adapt to a variable market around compute prices.

We are nowhere near market optimum for flop efficiency. That NVIDIA GPUs can appreciate in price is a leading indicator that real competition hasn't showed up yet. To the degree you are racing to purchase and hoard compute at long timelines puts you at tremendous risk.

I have never been busier over the past 4 years since GPT-4. There’s still seemingly zero risk of this for engineers, at least. This past month, I recently have found projects that take the normal 2-3 weeks again. Of course, these projects would’ve taken months. There’s a wide frontier out there.

I have never been busier over the past 4 years since GPT-4. There’s still seemingly zero risk of this for engineers, at least.

Looking for a good bike or bike shop to bike around SF (for me!). Any recommendations would be appreciated!

Looking for a good bike or bike shop to bike around SF with my son. Any recommendations would be appreciated!

The next quest after ASI is human life extension and thus immortality. Easily we should begin spending 10x a year on this as AI improves.

Very concerning situation

It does not seem like pacing or pausing can work. The only practical path is significantly accelerating. We must be so far ahead as a nation on both capabilities and alignment that by the time others catch up, our version of super intelligence can defend us.

Unfortunately, with access to an abliterated DeepSeek V4.1 model, there is little value to KYC holding back the frontier models now for cyber. In fact, making it hard to secure systems w extra friction in American Enterprise is hurtful to national security.

Tip. Of. Iceberg.

This is just the tip of the iceberg. New world.

If you're selling RL kernel optimisation tasks for models, pls dm me.

The greater the abstraction of what a company does for the longer the time without launching, the more likely it results in total failure.

I clicked on one post about a security incident and now my whole TL is filled with tweets about 5 new ones. Sheesh.

RT aidan We are proud to announce our series A(irplane). I wish I could share the full extent of my excitement for the incredible progress we've made, but there are not enough characters in the world. More details on our models and algorithms to come!

When you regrettably miss a good opportunity, remember you probably invented it and you can make more for yourself again.

RT immad Excited to announce @Mercury crossing $1bn/yr in revenue and 450k customers. Just getting started! https://x.com/i/article/2112209203666530304

🎯

are there any good aesthetic scorers out there for design?

If you can use models to reverse engineer anything and use models to swarm and find vulnerabilities, we are not ready for what's going to happen over the next 12 months. We've now have the autonomous black hat hacker. Tremendous crisis right in front of us.

One thing to consider is that the best hacks on American companies will go down very quietly. A great hacker is unseen. So you should presume that attacks on American companies that are noticed, are a small portion of what’s happening out there. Most hacks are just silent and steadily waiting to accumulate more valuable information on an on going basis.

Every organization in the world needs access to a strong frontier cyber model—ASAP. You should be sweeping your whole system as much as possible. This is the new world.

RT dax they've been non stop trying to attack our systems for the past 48 hours a side effect is everything is getting hardened but damn we basically need agents going 24/7 because that's what they have

Soham Parekh made it into the movie 😂🤣

🎯

Nathan Fielder is a mad genius. Incredible movie.

The agents swarms are coming. A tremendous amount of problems will be formulated into a compute problem.

I still think we vastly underestimate AI progress in the next 6-9 months despite the last 2 years. When I talk to people, I sense that it’s a function of possibly not wanting it to be true now. And so if your belief on this is to bet hard on the future, well, you sound delusional

> Simons Foundation Launches $15M Effort To Help Mathematicians Navigate AI Era. > "As artificial intelligence rapidly reshapes the research landscape, we want to help the mathematical community come together and forge the best path forward" Absolutely wild times.

What if we are in a RL env sandbox right now?

RT Daanish Khazi We are releasing Ultramarine 1 Flash, a frontier-class model built for knowledge work. It is open-source and achieves similar performance to Fable 5, while being 60x cheaper. It was post-trained by a team of 5 for ~$300k all-in. Available on API now, with weights releasing in 2 weeks.

Looking for a US based designer who is up for moonlighting a bit. Remote is ok. Need to have a few years of experience under their belt. DM.

I guess "the data guy" was good at being a "model guy." To say @alexandr_wang was every uni-modal at something was always quite silly—a great founder has to be polymath.

It is time to prepare for a world where there are tasks so complex and difficult to solve, humans will no longer know how they are solved.

I would love a Stripe Press of all the essays. @patrickc

🎯

Congrats @natolambert - excited for you. In the same way that I believe intelligence should be flattened and diffused broadly, the knowledge to build this technology should be as well. I can't think of anyone better!

The Vera Rubin era just booted up. 🫡

Dang, this dot thing really built Mighty into Codex. I’ve come full circle. The great cloud computer revolution has finally begun.

Get your .si domain people! Appreciate @rob at @snagged for helping us lock it down :)