Luke Kabbash

“And they may well ask, why climb the tallest mountain?”

@crushmarxismnow I want a @kalshi line for Tom Holland, Barry Keoghan or Timmothy Chalamet

Who are Pedro Pascal and Zendaya going to play?

@draecomino Can I help speedmog the TPUs?

I'm looking for a truly based technical product marketer. Must be fully steeped in the AI landscape and be a daily AI user/builder. You will get to market next-gen Cerebras models and hardware. Together we will speedmog GPUs. DM to apply. job-boards.greenhouse.io/cerebrassystem…

@AutismCapital Not mad about him going out like this. Mad the finale happened in a cardboard room, that they just *arrived* at the White House, that there was never a ‘snap’/blood & bone moment, etc

Not sure why people have a problem with Homelander going out like this. Convinced those people never actually watched the series. Homelander has always been a pathetic fragile ego coward that’s become increasingly unhinged and desperate. That’s the whole point of the character. This pathetic last ditch effort is perfectly on brand. Homelander was never going to die with dignity like a brave hero, he wasn’t a brave hero, he was a spoiled child. That was the whole point. This was realistic for the character.

@pashmerepat @kurorosage ‘Currently’ 😉

@kurorosage Similar, but codex currently lacks a few key things for a full-fledged personal assistant experience.

have a long running personal finance thread in the codex app set up with a heartbeat automation. as part of it, I gave Codex access to all my banks, credit cards, tax statements & brokerages. also built an agent-first Schwab CLI so it can view my holdings and open up trades. just asked codex to open source it. woke up this morning and saw this notification on my phone — codex made some trades (: at this point, codex knows my financial picture better than I do, and I use this pinned thread as my primary interface for anything money related. obviously handing over all your finances to ai has risks, but personally i’m stoked about this. feels good having codex in my corner helping me stay on top of frivolous spending, taxes, and portfolio

@MKBHD Bro is NOT feeling the AGI

This defeats the whole purpose of watching an unboxing video

@AlexBores Zero businessinsider.com/marc-andreesse…

This is the person trying to buy our Manhattan congressional district. (He's also one of Trump's top megadonors). We can't let him win.

@tokenbender @cain151714 Yeah obviously it’s just a search engine 😵‍💫 /s

clearly all the words and numbers used in this proof are already part of the training data. i went through the entire 125 pages of summarised cot, i couldn't find a single novel symbol in there. they really think they can continue fooling everyone with this. all the elementary pieces used to prove this were already known. not one new integer was invented. not one fresh Greek letter was discovered in the wild. the model simply rearranged previously existing mathematics into an order that happened to solve the problem, which is obviously just memorisation if you ignore the rearrangement, the dependency structure, the proof search, the synthesis, and the fact that this is exactly what mathematicians... ahem ahem anyhow, wake me up when it invents a completely new field all by itself and new greek letters and numbers with properties completely unknown. that is the true test for agi.

@browser_use Brilliant growth hacking holy shit

Use /goal wisely...

@NASA @NASAAdmin @NASAMoonBase Jamestown

We're building a Moon Base! @NASAMoonBase will serve as a habitat where astronauts live and work during long-term science missions. Join us at 2pm ET on Tuesday, May 26, for a live news event where we’ll share updates on our lunar exploration plans: go.nasa.gov/4uinkLi

EVE OS Agent can now make you fits, grounded in your skills and stats, directly importable directly in game. Gemini! Happy late April fool’s @HilmarVeigar 😊 eveos.space/fitlab/fit/d27…

@djcows @anduriltech Hey! It’s actually a quantized version of MiniMax 2.5, cmon!

@anduriltech finally, soldiers can ask qwen3.5 for advice during combat

Introducing Voyager Gateway 1. Warfighters at the edge need access to mission compute capabilities that are typically constrained to the base. The size, weight and power burden make server racks unrealistic options. Voyager G1 is the rugged, waterproof device that unlocks mission applications and AI-enabled workflows directly at the front line. It also connects operators as nodes on our Lattice Mesh network. This enables small teams to maintain operations even when networks are jammed, infrastructure is contested, or command posts are out of reach.

I feel like every one of these product surfaces should be consolidated into 1) Gemini app - Chat - Images - Music - Video - Flow - Stitch - Pomeii (? The marketing thing) 2) Antigravity/ai.dev - IDE - ‘Agent view’ - AI Studio 3) Antigravity CLI Anything new gets added to these surfaces. Antigravity should be able to call these other models (Omni, nano, Veo) via tool calling and have memory architecture capable of creating and saving design systems, characters, props, etc so the more ‘creative’ Gemini app stuff trickles down into the IDE. I feel like this isn’t rocket science and OpenAI was smart to get out ahead of narrowing their product surface killing Sora.

it’s in gemini, just create it in ai studio. oh, that’s for your personal google one account. for workspace you need gemini business. no, not gemini advanced, that’s ai pro now. unless you need ai ultra. oh agents? you do that in spark actually. no, not gemini api managed agents, that’s different. for coding use jules. unless you mean the agentic ide, that’s antigravity. no, that’s the old antigravity, download the new one. actually gemini cli is being deprecated, use antigravity cli. no the flash model is smarter than the pro model. unless you need pro. if it’s video, use flow. no, flow uses veo. no, nano banana is images. actually that’s in gemini now. unless you’re in search, then it’s ai mode. no, research is notebooklm. anyway it’s all very simple.

@OfficialLoganK Love it, but have some antigravity feedback x.com/lukekabbash/st…

Gemini 3.5 feels like the start of a new era for Gemini, we spent the last 2.5 years putting the infrastructure, products, team, etc in place (learning lots of lessons along the way). The model is the product, please keep the feedback coming!

@nikitabier can we just auto-mute people for posting like this -- I'm so tired. A setting for Premium accounts to auto-mute things that are >90% chance of being AI text would go so, so far. I know AI detectors are not flawless, and some well prompted Clauding will get through, but this slop has to go.

I continue to @gdb & @pmarca. I actually understand their incentives and logic - I think we need to win the AI (forever) race, even at a high cost. We also should protect our children, our vulnerable, and our poor. I feel like we can walk and chew gum here… but instead they’re giving 100s of millions to these organizations which are poorly run by absolute leeches suckling their money to make disingenuous YouTube ads… I know we can do better. The folks donating the big bucks should actually take some time and audit the content coming out of these slop-shops. If they want to be remotely well liked, this is not how to go about it. It’s also not secret, donating through a few proxies doesn’t make it any less obvious, and only serves to give you less control and empower the aforementioned leeches of your capital. (This isn’t @ you, Alex, it’s at the PACs and financiers)

NEW: I'm calling for a debate with the Trump-aligned super PAC spending millions to defeat our campaign.

@antigravity - 1) Lemme drag and drop projects to reorder them 2) Lemme edit sent prompts and resend from that point (or give option to not revert, but just modify context in flow) 3) When I hit '...' for a project, let me 'open in Finder' Some of the most important QOL is still missing! Love the app generally though. @OfficialLoganK

Lol someone had to make this cuz of me and I can’t even use Kimi on Cerebras If you’re gonna say ‘vs Cerebras’ just show the GPT-OSS number like you tried to dunk on me with 🔥

@cerebras x.com/lukekabbash/st… Don’t make me post the link again 🤦🏻‍♂️ It’s a *silly* hill to die on - why market it now? Just lemme buy Cerebras code again…

Google's new fast model 3.5 Flash vs Cerebras

@jmysct0 @draecomino @cerebras I don’t know the parameter size, I don’t believe it is public, but for 3.5 Flash vs Kimi K2.6:

@LukeKabbash @draecomino @cerebras It's not just TPS. Parameter size matters, so what is the metric for parameter size? Or "intelligence" of the model?

Oh look it’s $CBRS doing a little ✨ frame control ✨ Here’s Gemini 3.5 Flash running at 1,400 TPS on Google’s new TPUs (40% faster than Kimi 2.6 on @cerebras) youtube.com/live/wYSncx9zL… 726 P/E ratio btw 😉

@iamgingertrash @scaling01 All we are (productively) is - language - memory - opposable thumbs I used to be an LLM architecture doomer till I realized uh duh language (probably) is all you need, at least to automate the making of everything else

@scaling01 He’s so fucking stupid This is what happens to the European mind When the world progresses without them

Karpathy joins Anthropic to build RSI and Yann is still yapping about how LLMs are flawed