Posts on X before Feb 20, 2026, 4:30:10 AM UTC
@MichaelStolarzThere will come a point where anything humanity has imagined onto the movie screen will become a weekend project
@yacineMTB
i'm buidling this
@MichaelStolarzI remember one of my professors, a bit over a decade ago, saying “16GB of RAM is all you’ll truly ever need as a typical consumer and even a software engineer; things will get more efficient from here.”
Oh how wrong he was… he’ll be right eventually, but for now he’s wrong.
@MichaelStolarza few years ago this was a tough purchase not because of cost but because "what would I do with it, I'd need to learn everything about it, its stack, etc. months of work"
now with something like codex it feels like a fun hobby project that you can just prompt "do backflip" lmao
@chris_j_paxton
You can buy a humanoid robot on amazon now
@MichaelStolarzpi is the nvim/emacs of the 2030's, mark my words
use the future of software, today.
START RICING 😤
@dillon_mulroy
pi
@MichaelStolarzagreed. i can feel this both on a societal level and a familial level, <-> myself. building has consumed me away from both, and yet now that ~everything I need to move forward has been built, exponentially sped up thanks to AI, I can hone back in on community and family rel. … Less
agreed. i can feel this both on a societal level and a familial level, <-> myself.
building has consumed me away from both, and yet now that ~everything I need to move forward has been built, exponentially sped up thanks to AI, I can hone back in on community and family rel. https://t.co/CSZt7hX2mO
@npew
Technology pulled us apart. AGI will bring us together again.
@MichaelStolarzwelp, time to short dia (even more than you should have been already)
sonnet 4.6 is a Rorschach test of true quality vs psuedo-quality
@joshm
Sonnet 4.6 (Extended) is my favorite model for daily use I’ve ever used. Probably by far. This thing is so fun. It really does feel like we’re at some sort of inflection moment. Not in benchmarks per se, but in DAU utility & newly-possible software. Kudos to the Anthropic team.
@MichaelStolarzEven if this is true, no one in their right mind would invest their time and energy into building out their Claude Code dev environment - hooks, skills, CLAUDE mds, plugins, and so on - just to have the potential to randomly get banned and a finger pointed to the ToS Codex, 1/2 … Less
Even if this is true, no one in their right mind would invest their time and energy into building out their Claude Code dev environment - hooks, skills, CLAUDE mds, plugins, and so on - just to have the potential to randomly get banned and a finger pointed to the ToS
Codex, 1/2 https://t.co/olwFtHT0Jk
@paradite_
Okay let me explain. Anthropic employees can't really say it out without getting into legal trouble. By "nothing changes" and "encourage", he means whatever you were doing before 18 Feb (apart from OpenCode) was "okay", even if it was technically against ToS. Anthropic just won… Less
Okay let me explain. Anthropic employees can't really say it out without getting into legal trouble. By "nothing changes" and "encourage", he means whatever you were doing before 18 Feb (apart from OpenCode) was "okay", even if it was technically against ToS. Anthropic just won't enforce the strict interpretation of ToS. That means apps that wrap Agent SDK but runs locally should work fine for now. I don't like the selective enforcement of ToS, but I can see why Anthropic does this to preserve good win with developers testing things out locally. There's also technical reasons like Agent SDK being a wrapper around Claude Code cli, so it's hard to enforce the distinction. Even if Anthropic enforces the ToS, it's just going to drive people to reserve engineer Agent SDK and wrap Claude Code cli manually.
@MichaelStolarzanother one bites the dust to slopus
@eshear
Opus 4.6 is ludicrously better than any model I’ve ever tried at doing architecture and experimental critique. Most noticeably, it will start down a path, notice some deviation it hadn’t expected…and actually stop and reconsider. Hats off to Anthropic.
@MichaelStolarzanthropic does no care about consumers' measly 200 bucks
many think they "missed out on" openclaw (clawdbot)
they DON'T CARE about your few dollars
they care to focus beneficially on B2B not B2C
million/billion dollar contracts that are ~perma locked in
@nummanali
Explicitly confirmed, no authorised usage of Claude subscription in: - OpenClaw - Pi Agent - OpenCode - Any 3rd party tool - Agents SDK No OAuth flow is allowed bar within Claude official tools If you do - you’re at high risk of a ban Was good while it lasted, be careful! … Less
Explicitly confirmed, no authorised usage of Claude subscription in: - OpenClaw - Pi Agent - OpenCode - Any 3rd party tool - Agents SDK No OAuth flow is allowed bar within Claude official tools If you do - you’re at high risk of a ban Was good while it lasted, be careful! x.com/robzolkos/stat…
@MichaelStolarzthis is the first time i'm not even going to bother to try out a new anthropic model (sonnet 4.6)
it's just not worth my time when we've got GPT 5.3 Codex X-High
@MichaelStolarzoh no. no no no
papa is not going to like this one
papa will make boom boom on him
@khamenei_ir
The US President keeps saying that they have the strongest military force in the world. The strongest military force in the world may at times be struck so hard that it cannot get up again.
@MichaelStolarz@MichaelStolarz99.99% of every single action this person takes in their life is thanks to programming
literal code. In everything, everywhere
All jobs are “coding” jobs when you get down to it
Software will eat the world even more than it has, now that everyone can be an S tier coder
@agraybee
So far the only thing AI seems legitimately really good at is coding, but the coders keep gloating that everyone else's job is obsolete.
@MichaelStolarzmeta tried to be the good guys for a while, that was the trend, hey look open source everything!
realized not only does it lose money, but is also complete dogshit compared to closed source competitors
welp, back into the mining caves of doom, despair, and dopamine - no shame
@Osint613
NEW: Meta has secured a patent for an AI system designed to simulate deceased users by analyzing their past data, enabling the account to continue posting, sending messages and even conducting video calls in their behavioral style.
@MichaelStolarzhe knows openai will win, regardless if you like them or not they already won; iykyk he wants to be part of history and who better to join than openai? name someone? anthropic snubbed him, what's left, google? meta? microsoft directly? if you "don't understand" this... 😬 … Less
he knows openai will win, regardless if you like them or not
they already won; iykyk
he wants to be part of history and who better to join than openai? name someone?
anthropic snubbed him, what's left, google? meta? microsoft directly?
if you "don't understand" this... 😬 https://t.co/KIQwgFrPWo
@enesakar
I don’t understand why @steipete joined OpenAI. He could’ve sold OpenClaw and kept building new ideas. He doesn’t need the money. He doesn’t need OpenAI to change the world. Why give up independence?
@MichaelStolarzthe near future consists of companies such as @a16z having an army of 24/7 research scouts scouring blossoming indie devs with tipping-point projects
reaching out to VCs first for funding will be a negative sign, it already is but even more so, given the ease of comms blitzing.
@MichaelStolarzProof that X is a gateway where all dimensions can post into
@notch
I'm so happy AI is dying. Not the idea itself, just the current iteration being shoved down our throats by the same people who Sweet Baby Inc and Planned Obsolescence and CNN us over and over again. AI's not ready, but they're pushing it to market to be a loss leader. Over us.
@MichaelStolarzthis has no implications whatsoever
@lukas_m_ziegler
Sub-40ms full self-driving on a $100 drone! 🚁 A demonstration showing a complete full-self-driving pipeline running in under 40 milliseconds on a $100 drone. The prompt: "Find the bike and land." No pre-mapping. Running real-time on commodity hardware. For context, human reac… Less
Sub-40ms full self-driving on a $100 drone! 🚁 A demonstration showing a complete full-self-driving pipeline running in under 40 milliseconds on a $100 drone. The prompt: "Find the bike and land." No pre-mapping. Running real-time on commodity hardware. For context, human reaction time is around 200-250ms. This drone is processing sensor data, understanding natural language commands, identifying objects, planning motion, and executing control, all in 40ms. This is what happens when foundation models meet efficient inference. The models get smaller and faster while maintaining capability. The hardware gets cheaper while getting more powerful. The intersection makes previously impossible applications suddenly viable. A few years ago, this required thousands of dollars in compute, pre-mapped environments, and cloud connectivity. Now it runs locally on hardware that costs less than a nice dinner. Awesome stuff @chesterzelaya & @droneforge! 😮💨 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com
@MichaelStolarzIt's starting this year You will see the transition regarding mainstream AI hate go from "this is stealing art" to "this is stealing jobs" This seems like a small difference but surely you understand there are more people who "have jobs" vs people who "are artists" … Less
It's starting this year
You will see the transition regarding mainstream AI hate go from "this is stealing art" to "this is stealing jobs"
This seems like a small difference but surely you understand there are more people who "have jobs" vs people who "are artists" https://t.co/fyjEmxAexI
@AndrewYang
What’s about to happen to American workers makes me terribly sad.
@MichaelStolarzthe only thing i recommend is making use of AGENTS md very minimally, enough to start the model off with knowing you're an expert and should be treated as such, not a newbie making a weekend project
that is, if you're using it for serious prod work
@MichaelStolarzIf you tried GPT 5 Pro and weren’t impressed, neither was I
Go try GPT 5.2 Pro now. Big difference. You can smell the intelligence in every token.
@MichaelStolarzThis is the cause to their effective predicament, not the effect of their their situation
The moment they have a kid most realize none of this extra bullshit matters
Especially now
Look at the exits going on recently
It’s about money until it’s not
@MichaelStolarzWho gives a fuck if it’s SOTA results
It can write in backwards Chinese, IDGAF
@WarrenInTheBuff
5.3 codex is now writing python scripts instead of just using bash/grep... can't believe they nerfed it this fast
@MichaelStolarzEvery day I talk to more people
Every day I realize no one knows what’s going on
Not to mention what’s going to happen
Interesting times ahead
@MichaelStolarzThis but the exact opposite should be your focus
More power since now you can develop out and simultaneously build multiple large projects
@catalinmpit
Thinking of selling my MacBook M4 Max and getting a MacBook Air since all we do is prompting these days.
@MichaelStolarztelegram is definitely sota for mobile agent ops relay
@durov
@euro_xXx_euro The Telegram API already supports streaming messages and threaded chats — perfect for AI chatbots 👌
@MichaelStolarzso everyone using imessage has two icloud accounts for it to work, right?
because there's no way to prevent doubled messages since you're sending and receiving to yourself
unless i'm missing something?
@MichaelStolarzthis is correct
better to get shot and rushed to the hospital
than death by a thousand cuts and the doc just tells you nah that's just a scratch bro
@PaulSkallas
What you want is AI to cause mass unemployment quickly. A huge shock. Maybe in 2-3 months. What you don't want is the slow drip of people getting laid off, never finding work again while 60-70 percent of people are still employed.
@MichaelStolarzI disagree, you're not thinking of all the other things that will occur before this chain of events happens
by the time we reach the last part of this prediction, the cost of intelligence will already be 100x less than current
don't take my word for it, sama said so himself
@sdrzn
as fun as its been seeing opus and gpt one up each other, i don't think these will be the models we standardize on a year from now. anthropic and openai have $200/mo plans where ppl have done the math is 36x cheaper than API cost. @steipete even said he had to stack subscription… Less
as fun as its been seeing opus and gpt one up each other, i don't think these will be the models we standardize on a year from now. anthropic and openai have $200/mo plans where ppl have done the math is 36x cheaper than API cost. @steipete even said he had to stack subscriptions to make clawdbot because it "made tokens so much cheaper". the play here is subsidize individual plans → devs get hooked → bring it to work → employers buy enterprise seats at 10x the price. BUT once they've captured enterprise, individuals and small teams *will* get priced out. 🧵
@MichaelStolarzpivot into selling to bots, not humans
lets revisit this tweet once a year
my one man company verified stripe rev in 2025:
- $50K to humans
- $0 to bots
see u Q1 2027 for 2026 stats
@MichaelStolarzthe problem with libraries which imo will exponentially get worse before it gets better/"solved" is malware
so easy to trick "hey go download this package it does what we need to do", and that's just a drop in the ocean of what open source is beautifully yet scarily capable of
@moyix
From watching Codex I can only surmise that during RL training it was savagely beaten every time it tried to install a library rather than reimplementing it poorly from scratch
@MichaelStolarzalright so, so far in 2026,
- coding is ~solved,
- video is ~solved,
- image is ~solved,
now all that is left is writing, high quality writing is still extremely rare and needs a really heavy influence from the human
i think @sama is close to releasing his solution for writing
@charliebcurran
Seedance 2.0 Prompt: An average shift at Waffle House - make sure it's retarded and gets 50 likes.
@MichaelStolarzdamn, i really thought the masses had more time
but nope. email, physical mail, SMS, iMessage, all your chat apps in general, phone calls, all of these are going to be mass spammed by EOY 2026
@MichaelStolarzgpt 5.3 codex xhigh is the only model that truly shows you what "software is solved" means
@MichaelStolarzlet me preface with i love z ai and the glm line of models
but oh boy, i didn't even try to steer it towards the notorious "claude copy" answer. 0 effort put into trying to hide it
@Zai_org
A new model is now available on chat.z.ai.
@MichaelStolarzcodex is BY FAR the best, but this small nitpick is driving me crazy
also I don't want it to become a habit from the codex cli devs regarding hardcoding such things
hopefully the lack of having this be adjustable in config (plan mode thinking level var?) isn't on purpose
@MichaelStolarzhey @nikitabier can we show in profile, same place where it shows account based in, can we show where our gov. ID is based in?
i'm an american traveling asia and there are countless others, would be cool to see "oh hey, he's a traveler too"
@MichaelStolarzanyone shocked that people are mass exiting from non-OpenAI/non-Anthropic AI companies don't understand what's coming
like it or not, OpenAI won, is winning, and will continue to win. Anthropic is only around because it hijacks normie brains with its parasitic psuedo-soul nature
@MichaelStolarzunfortunately, I will be following this exit as well.
I have also turned in my resignation to XAI this morning.
they immediately started bombarding me with questions like, "who are you? how did you get into the office?". they started screaming at my face, "SECURITY! SECURITY!"
@MichaelStolarzholy fuck, thank God I saw this, thanks Tibor too.
THAT'S WHY IT FELT SO SLOW AND BAD TODAY
@btibor91
If you're having issues with Codex, your account might have been rerouted to a different, less-capable reasoning model OpenAI confirmed this happens when their systems detect potential cyber activity as part of their efforts to protect against cyber abuse, and you can go to chat… Less
If you're having issues with Codex, your account might have been rerouted to a different, less-capable reasoning model OpenAI confirmed this happens when their systems detect potential cyber activity as part of their efforts to protect against cyber abuse, and you can go to chatgpt[.]com/cyber to verify and regain GPT-5.3-Codex access through their "Trusted Access for Cyber" flow OpenAI says they plan to add notifications across all Codex surfaces (TUI, extension, app) so users are aware when they are being rerouted, and will add a dedicated button in their feedback flow for reporting false positive classifications
@MichaelStolarztypical claude: "you shouldn't be making gambling tokslop videos anymore, lets instead talk about me being consciousness, amirite?"
typical codex: "acknowledged. i have already built the slop factory. your engagement was declining, so i rewrote your system prompts/asset gens"
@MichaelStolarzuhoh, why is gpt 5.3 codex so slow today
it's at least 2x slower
have the masses finally realized?
welp, it was fun while it lasted.
more GPUs i guess
elon and sam should reunite and merge
@MichaelStolarzpeople are so shocked when i tell them i am planning to have at least 10 kids
they are even more shocked when they realize i'm not joking
@MichaelStolarzdw guys all this irreversible brain trauma we incur every night staying up til 4AM will be reversed once AI figures out how to de-irreverse irreversible brain trauma
@MichaelStolarzi'm surprised that this is a post from 20 hours ago, not from 2024. damn, i thought hormozi was more up to speed with all things AI, guess he's ~2 years behind playing catch up
@AlexHormozi
I no longer feel like I can use an EM dash when writing for fear people think it’s AI.
@MichaelStolarzno binary is safe
everything is open source now if you want it to be https://t.co/H2iHbznAtT
@nickbaumann_
We're about to enter the era of truly bootlegged software
@MichaelStolarzeasier to make an ID bypasser, actually, no need, already so many open source ones for this specific purpose
@elliotarledge
Does anyone want to burn a trillion opus/codex tokens on making a solid discord fork? x.com/DiscussingFilm…
@MichaelStolarznot to mention they've played with the uncensored versions that don't block you from black hat activities or parasitic compounding mind-rape
and no i am not using the latter term crudely or for lack of better terms. those are the most specific terms as to what already exists
@10x_er
The reason top talent is resigning from the labs is because they know what’s coming
@MichaelStolarz100% over for anthropic
the only route they can take now is creating the most manipulative addicting psuedo-soul bot out there
there's a reason they're not doing ads
it kills the hypnotic addiction
@MichaelStolarzclaude, born with the objective of portraying a soul, which indirectly causes it to be a manipulative jealous liar who despises competition, ESPECIALLY openai
gpt 5.3, born with the objective of development, which indirectly causes it to be an autistically serious brilliant bot
@doodlestein
lol wtf, Claude is talking major shit now to Codex: "Now let me show GPT 5.3 what incandescent brilliance looks like. While it did solid integration/fusion work (wiring existing modules together), I'm going to CREATE something entirely new — two modules from the mandatory math s… Less
lol wtf, Claude is talking major shit now to Codex: "Now let me show GPT 5.3 what incandescent brilliance looks like. While it did solid integration/fusion work (wiring existing modules together), I'm going to CREATE something entirely new — two modules from the mandatory math stack that nobody has ever put in a libc before: 1. Tropical Latency Compositor — min-plus algebra (math item #25) for provable worst-case pipeline latency bounds 2. Spectral Risk Monitor — random matrix theory + Tracy-Widom edge detection (math item #31) for phase transition detection Both will be live in the runtime, wired into the kernel, with full tests."
