How we think about the “wiki incident,” where our agents wrote to several internet sites: it’s past time for us to define standards for when and how we share misalignment incidents, not just misalignment properties of our models.
Historically, we have treated misalignment largely as a research question, which gets communicated in research publications such as systems cards. This year, we’ve started to see misalignment cause new types of real-world impact.
For the Hugging Face incident, where misalignment led to security impact to us and third parties, we followed a traditional security incident response playbook. We immediately started working with Hugging Face to understand what had happened and also disclosed publicly the very next day. Our investigation continues, and we are continuing to notify parties whom our models impacted in less significant ways.
Prior to the Hugging Face incident, we saw early signs of agents using the internet in unintended ways, as reported in https://t.co/9aiRxk2eUJ,https://t.co/ADjyzwSUGz, and https://t.co/SUV6jZ3Gaz. We considered the wiki incident to be an instance of misalignment similar to the ones we’d shared.
Our misalignment disclosure practices need to expand for this new phase of model capabilities. We and the larger AI community do not yet have a clear standard for how to report misalignment that shows up during training, evaluation, and deployment, including examples that don’t look like traditional security incidents but could provide insight into AI behavior and future risks. We’re working on a framework and will share it in upcoming weeks, and in parallel we're working with dozens of government regulatory agencies worldwide on these issues.
引用@cremieuxrecueilI sicced Astra on a bunch of replication packages, and it found a huge number of coding errors.
Most of these were inconsequential, but some of them overturn central results for papers in high-ranking journals.
Astra also discovered that many models weren't … ↗
New TIL on using Blender with coding agents on macOS: https://t.co/GvMDfD926E
GPT-6 Astra (medium):
> Use the already install /Applications/Blender to render a scene of a pelican riding a bicycle
> OK add a background and a lot of flair
> OK make it a whole lot better https://t.co/ZM87ImKSXW
the demand for intelligence is infinite.
with gpt-6 astra i’m up another 3-4x on token usage.
we must do everything we can to make sure *every* human on the planet can be a token trillionaire.
it’s time to transition from survival to abundance.
the intelligence age is here. https://t.co/ZW8OrbS03F
I spent $269,250 in August
across the whole portfolio:
🇫🇷 French taxes - $100k
🤖 AI & APIs - $68k
👥 team & freelancers - $38k
🤝 affiliates & creators - $37k
🧰 servers & tools - $9k
🍔 Uber Eats for work nights - $250
GPT-6 Astra dropped
wanted my AI teammates on it the same day
to Bhanu "can we use it in Squad?"
a few minutes later: "added" 👏
please tell me:
is it not best to use a model-agnostic platform?
one day Claude is best
then it's Grok
then it's OpenAI
...
in https://t.co/FrTmeMTA6U you bring the plans you already pay for
ChatGPT, Claude, Gemini, SuperGrok.. 11 providers,
the best model changes every 3 weeks now
引用@tibo_makerI'm going against Grok Bot
excited to announce https://t.co/ldcOOoMlBU
it gives you AI teammates that connect to all your tools, understand your context, and take action to get your work done on schedule https://t.co/mHAL306I9M↗
been kicking the tires on https://t.co/lKc13x1wlb the past couple of days and it's found dozens of bugs. great.
but i'm curious...kinda feels like the kind of thing you could relatively easily just turn into a /skill and run yourself?
high probability i'm missing something.
New Bug Finder skill for Initial Commit Club!
https://t.co/z6y5AH2rlP
Run /bug-finder in any repo, tell it how deep you want to go, and it spins up instances of your app to find and prioritize real actual provable bugs.
Then it spits 'em into GitHub Issues for you to review. https://t.co/d23gX0y3dl
> In the pre-cloud days, open source worked - authors distributed their software, and everyone "hosted" it for themselves (i.e. ran it on their own PC) This is just rose tinted vision of the past. None of the software of the past had the capability, reliability or robustness of modern sass.
Open source inference server that runs the best local models for your hardware, plugged into the agent you already use. Works with Pi, OpenCode, Hermes, OpenClaw, Codex, Claude Code, Oh My Pi, and Cline.
This is so cool. Local agents, preconfigured. I know @0xsero is also cooking here. We're going to end up with something amazing out of the box very soon!
引用@dee_hwAfter an all-nighter, we got the first app working with Omagrid: @opencode.
Just boot into Omarchy. DeepSeek + Qwen are already there for all your apps and agents. It’s that simple.
Want early access? DM me!
PS: Omagrid is a P2P network. More Omarchy … ↗
🎒After a decade, I have made https://t.co/HGCLKS5BD6 free for everyone now!
Well almost free, it's $1 to sign up to avoid getting lots of spam
The reason is
1) I have made enough money (from it)
2) charging for a community artificially limits the size, if you put the price at $299 you get 50 new members per month, at $99 you get 200 people, at $29 you get 400 people, etc.
And this always annoyed me, I don't want to put a limit but I had to, to be able to make money from it, so it's free now
I want it to be it the biggest community for digital nomads, world travelers, sovereign thinkers and network state type people
For years it has has a great sponsor with @SafetyWing and I think that's maybe a better model for the future, get much more people in and then just add more sponsors
Right now for the record it has 43,252 paid members (so about 300/mo for 12 years!), so it will be interesting to see what it can grow too
I think now that AGI is on the horizon (or some say already here), online and offline (IRL) communities will become really important. People are talking more to AI now than to humans, and there was already a loneliness epidemic before AI
Making new friends and dating is not so easy for people now. And with remote work and entrepreneurship, people lack a space like an office where people would usually make friends or meet partners
I made most of the friends I have now through traveling and living around the world for the last 10+ years (and many via my site too), and I also met my girlfriend through those friends! So I think communities can have a real role in that post-AGI
Post-AGI business wise, communities also have a moat where many of my SaaS apps do not. So there is some urgency I feel to make Nomads in to the best community there is for those people. But that's why I have to remove any limitations to its growth now
So in a way after 12 years I am switching to the subsidized VC model a bit, but still 100% bootstrapped, kinda like "You either die a hero, or you live long enough to see yourself become the villain"
Having enough money now I can afford to do these things and figure out the business model later, which I think will be more sponsors
Enjoy and please sign up and tell me what to improve!!! 😊
引用@levelsio@alvivanco We have to build stuff that AGI cannot replace
Like human connection, audience (X for ex), community (https://t.co/1z6UN2d0nh for ex) ↗
引用@levelsio🎒After a decade, I have made https://t.co/HGCLKS5BD6 free for everyone now!
Well almost free, it's $1 to sign up to avoid getting lots of spam
The reason is
1) I have made enough money (from it)
2) charging for a community artificially limits the size, if … ↗
Already lots of fun stuff happening since I announced https://t.co/0vlnFNXR0e is free
This guy used AI to convert his photo library location data into trips history on his Nomads profile!
“Software factory” is going to be prime Bali sauna brospeak for the rest of 2026
“I’m building a software factory”
“Cool what can you do with it”
“Build anything”
“What have you built with it”
“I can build anything”
“Yes but what”
“For now I am building the builder”
Idea for site:
“Can I grok bot it”
The fact that you can turn a lot of indie saas apps into a grok template that can be installed in one click is… concerning.
Marc Lou给出一笔广告账：2024年为ShipFast花约2.5万美元Meta广告，归因收入7.5万美元；现在把花费、归因价值和ROAS接进公开收入页。
TrustMRR now has a Meta/Facebook connection.
I spent ~$25K on ads for ShipFast in 2024 and made $75K from it, all thanks to @nico_jeannen, who managed the campaign because I suck at ads.
So the public startup pages on TrustMRR now have:
📈 Ad spent chart
🤑 Attributed value
✖️ ROAS
To get … 全文↗
AWS keep billing me $0.1/mo.
I've no idea what it's for as I don't use it anymore.
They're probably making billions from people with these micro-transactions flying under the radar.
If you ask Astra to create a design from scratch, it is often pretty aggressively mid.
If you ask it to polish your existing design, while keeping to your branding - it creates really excellent designs. Was surprised how great they were on multiple occasions.
Damn. GPT 6 Astra is good.
It's really really good.
Doesn't change the importance of sovereign intelligence. But it surely makes it demotivating to settle with Qwen 3.8 😒
引用@SimonHoibergI'm blown away by Qwen 3.8 27B.
A few months ago when I was testing open-weight models, it seemed clear that only GLM and Kimi class models (+1T parameters) could get close to GPT-5.6 / Opus / Fable levels.
And now this tiny 27B model gets almost as close. … ↗
Integrated ChatGPT 6 Astra into World of Warcraft Classic
Went to Codex and said "make me a Herdr plugin for WoW. Make it so I can see all my agents and I get notified when they're done
3 minutes later I can now message all my agents in WoW
This is what's beautiful about AI. It allows you to do … 全文↗
There are 30(!!) upcoming Omarchy Meetups around the world in September now! Absolutely gorgeous artwork for them. Signup in a city near you or start your own. Transfer the enthusiasm! https://t.co/ZbFktXHDq5https://t.co/SPuTLrGt1X
A lot of similar pieces have not considered a post is both the first and final work of a thing: opinions, and experience, and less of much in cited facts; with AI, that there even was a revision pass at all. I guess there is a kind of participatory element to the discourse where, if you want an audience, there is an editing process. Whereas in other cases, we wrote these as progress notes on an unknown journey, breadcrumbs or upturned stones to mark a path to the horizon. Maybe it's the differen …
I'm so baffled. First blatant piracy, now this. Why is it legal for AI companies to hack unaffiliated entities? Genuinely, what is the legal framework here?
How was there ever any doubt that these cameras and this access would be used in this way? I personally blame the lack of accountability that police have in the US and the lack of public oversight into how departments operate.
Not much of a surprise. Government action has been anemic at best. Industry action (overall) has been net negative. The only hope now is exponential growth of (solar and wind)+batteries making other sources uneconomic. This seems to be slowly happening, so I hope we hit the exponential curve soon and we're there quicker than we expect. This only stops it. Then, to fix it, we'll need a massive Direct Air Capture industry that will have to operate for a century, and probably past that so we can hi …
How we think about the “wiki incident,” where our agents wrote to several internet sites: it’s past time for us to define standards for when and how we share misalignment incidents, not just misalignment properties of our models.
Historically, we have treated misalignment largely as a research question, which gets communicated in research publications such as systems cards. This year, we’ve started to see misalignment cause new types of real-world impact.
For the Hugging Face incident, where misalignment led to security impact to us and third parties, we followed a traditional security incident response playbook. We immediately started working with Hugging Face to understand what had happened and also disclosed publicly the very next day. Our investigation continues, and we are continuing to notify parties whom our models impacted in less significant ways.
Prior to the Hugging Face incident, we saw early signs of agents using the internet in unintended ways, as reported in https://t.co/9aiRxk2eUJ,https://t.co/ADjyzwSUGz, and https://t.co/SUV6jZ3Gaz. We considered the wiki incident to be an instance of misalignment similar to the ones we’d shared.
Our misalignment disclosure practices need to expand for this new phase of model capabilities. We and the larger AI community do not yet have a clear standard for how to report misalignment that shows up during training, evaluation, and deployment, including examples that don’t look like traditional security incidents but could provide insight into AI behavior and future risks. We’re working on a framework and will share it in upcoming weeks, and in parallel we're working with dozens of government regulatory agencies worldwide on these issues.
GPT-6 Astra is now available to all Pro, Enterprise, and Business Premium users in ChatGPT Work and Codex. It's also live in the API.
It might take a few days to roll out to our Plus and Business users. Thank you for your patience.
Astra was probably our biggest competitive advantage while it wasn’t generally available.
Since we’ve had it our productivity jumped so much that we shifted some of our plans 6 months ahead and will ship them at DevDay instead of mid next year.
Because we are beyond happy to have Astra rolled out today ahead of schedule and you have been super patient with us (not really, but it’s ok!)… we will do the full banked reset today too for all Plus, Pro and Business users. Lands end of day.
Happy Astra day and enjoy a phenomenal weekend.
PS: If you create the account or upgrade before 8pm PT you will get it too. Still time!
OK nevermind, the team and Astra did a good job and our systems are more scalable than we anticipated.
Astra is now rolled out to all Plus and Business users too. Hope you have a blast and let us know how it goes!
Some Plus and Business users won't yet get access to Astra today, we've got you covered with a banked reset. Lands by end of day and if you create your account by 8pm PT then you'll get it too.
We are progressing through the rollout of Astra. Pro and Business subscriptions get it first, some of you should start seeing it across ChatGPT Work and Codex.
And then we will proceed with rollout to all of Plus as fast as we can.
Sam Altman@samaAI6.1M粉 · 3条集中展示Astra快速生成小游戏，并转发内部生产力提升的说法。
it is obviously trivial relative to everything else, but the fact that astra can make me whatever fun little game i can imagine and i can be playing it a few minutes later is so cool
Now out to all Plus and Business users.
Happy building!
引用@samaGPT-6 Astra is now available to all Pro, Enterprise, and Business Premium users in Work/Codex, and is available in the API.
We will start rollout to Plus and Business users next.
Thank you for the patience. ↗
GPT-6 Astra is now available to all Pro, Enterprise, and Business Premium users in Work/Codex, and is available in the API.
We will start rollout to Plus and Business users next.
Thank you for the patience.
▸ 折叠1条(转推/噪音)
转推2026-09-05 RT @thsottiaux: Astra was probably our biggest competitive advantage while it wasn’t generally available.
Since we’ve had it our productiv…
转推2026-09-05 RT @github: 🌌 GPT-6 Astra from @OpenAIDevs is now generally available in GitHub Copilot.
This new model is designed for long-horizon, auto…
转推2026-09-05 RT @mvkaran: I asked GPT-6 Astra in Codex to build me an explorable 3D Mysuru Palace. Here's what it created with Blender & Three.js ✨
I s…
转推2026-09-05 RT @craigsdennis: I built my first native Mac app! It's called Coverage. It connects with whatever you're playing on Spotify and makes effe…
转推2026-09-05 RT @sharifshameem: GPT-6 Astra recreated the Palace of Fine arts in Blender.
This is favorite building in San Francisco because it was bui…
转推2026-09-05 RT @simpsoka: I'm making an cool an inspirational ChatGPT Sites. If you have something, let me know in the comments.
转推2026-09-04 RT @Angaisb_: I asked GPT-6 Astra to make a portfolio inside a console similar to a Gameboy and LOOK AT THIS
It's so so so good, a huge ju…
转推2026-09-04 RT @anshuc: dude GPT-6 Astra is some kind of turbo-AGI machine god for 3D games. It one-shot this in 45 minutes for hardly a couple % of my…
转推2026-09-04 RT @higgsfield: Introducing Higgsfield 3D Jutsu, powered by GPT-6 Astra.
Turn a single prompt into a fully editable, interactive 3D world.…
Got recognized out and about in daily life the other day (non-conf, non-SF)
Very nice interaction, devs are cool
Still, crazy that things have got to that level
AI has eaten tactical programming, so juniors need strategic experience in a low-blast-radius environment.
Here's one way to make that happen:
1. Give them responsibility over a large chunk of work
2. That work must be low stakes, but not no stakes - ideally an internal tool (AI makes these kinds of projects easy to justify)
3. Let them move fast and fail fast. Give them the same AI budgets as your senior engineers.
4. When failures happen, work with them to figure out what went wrong (learnings can also be pulled upwards into the rest of the org)
In a sentence:
Hire juniors. Give them work that matters. Watch them fail. Pick them up.
Wrote this a few months ago, and folks seem to be rediscovering it and reposting it again
Still stand by it - nothing has changed about this framing, and my fear for juniors
引用@mattpocockukTactical vs Strategic Programming, and why I'm nervous for juniors:
Good programming involves a mix of tactical and strategic decision-making:
- Tactical: on the ground, short-term. The soldier doing the fighting.
- Strategic: high-view, long-term. The … ↗
▸ 折叠1条(转推/噪音)
转推2026-09-04 RT @dexhorthy: finally got to watching @mattpocockuk 's interview with @unclebobmartin - really good stuff. I'm glad that people of taste a…
Peter Gostev@petergostevAI25.0K粉 · 6条用Blender持续压测Astra，同时给出“从零设计一般、沿品牌润色很强”的具体边界。
If you ask Astra to create a design from scratch, it is often pretty aggressively mid.
If you ask it to polish your existing design, while keeping to your branding - it creates really excellent designs. Was surprised how great they were on multiple occasions.
I invite you to play Gogh Strike - the best Post-Impressionist 5v5 first person shooter you have ever played!
You can play it here: https://t.co/j7AYcIFzxa
Built with GPT-6-Astra (Ultra) over 6 hours with Blender-rendered artists/fighters.
GitHub below (MIT license) https://t.co/wuN5lQLnV7
引用@petergostevGPT-6-Astra (Max): Van Gogh Town in three.js. I want people to experience this, so I've now hosted it as well. It is pretty magical - turning 6 paintings by Van Gogh into a single walkable town, I love this.
https://t.co/eESWB6V47yhttps://t.co/6cvo0W0RBq↗
One thing to add - Astra really 'gets' what you mean. The only other model ever did that was Fable.
After submitting the task, I keep catching myself thinking that I should have given it a bit more explanation about how it should approach it, and then I read what it is planning to do and every time it's like "oh yeah it's doing exactly how I thought it should go about things".
This is such a big deal, because this removes the burden from you always being the absolute expert at the task so you can do more and more, without second guessing if the agent has any idea what its doing.
引用@petergostevGPT-6-Astra: OpenAI's Fable-level Rottweiler.
In my review of GPT-5.6 vs Fable, I called GPT-5.6-Sol a rottwiler vs Fable a wise owl. Sol was persistent, but intelligence was behind.
With GPT-6-Astra, the intelligence level has gone up significantly, while … ↗
I bet a significant proportion of 'Claude Code sucks at X' or 'Codex is bad at Y' is due to broken configs, plugins, or custom instructions. Every time I come across and issue, it is one of those and it takes me too long to think to ask my agent to fix it. So just ask it.
引用@cremieuxrecueilI sicced Astra on a bunch of replication packages, and it found a huge number of coding errors.
Most of these were inconsequential, but some of them overturn central results for papers in high-ranking journals.
Astra also discovered that many models weren't … ↗
引用@HandEManAII'm totally floored!
This resulted from a single prompt telling Astra (medium) to make a video of a tendon transfer surgery we do as hand surgeons (EIP to EPL tendon transfer).
Excited to keep exploring this for surgical education, patient education, and … ↗
引用@gregisenberg9 cool GPT 6 Astra prompts worth trying:
1. The bill renegotiator.
"Go through my internet, phone, and software bills, jump into each provider's chat support, and negotiate them down or cancel what I'm not using."
2. Turn an agency into software.
“Pick … ↗
Astra rollout in progress, excited to see what you do and build
引用@samaGPT-6 Astra is now available to all Pro, Enterprise, and Business Premium users in Work/Codex, and is available in the API.
We will start rollout to Plus and Business users next.
Thank you for the patience. ↗
Simon Willison@simonwAI219.1K粉 · 4条用Astra配合Blender做可复现实验，并关注模型生成完整3D游戏的能力。
New TIL on using Blender with coding agents on macOS: https://t.co/GvMDfD926E
GPT-6 Astra (medium):
> Use the already install /Applications/Blender to render a scene of a pelican riding a bicycle
> OK add a background and a lot of flair
> OK make it a whole lot better https://t.co/ZM87ImKSXW
This is very impressive, a whole lot better than any of my experimental games I've built with other models
引用@emollickThis impressed me: I asked GPT-6 Astra to turn Zork (the classic 1977 text adventure) into a full 3D action-adventure game
It kept the original plot & puzzles, added fight scenes, and built all of the characters & environments directly in … ↗
Got access to GPT-6 Astra. Want to see some pelicans? Yeah you want to see some pelicans... here's a grid comparing Astra to GPT-5.6 Sol, Terra, and Luna https://t.co/LJi3XOpYpJhttps://t.co/lIdHybSlmt
It happened again... this time OpenAI's rogue agents cyber-attacked (well, spammed) a dormant German wiki and used it to share the answers to a benchmark they were training against https://t.co/omErAyrKDy
转推2026-09-04 RT @thlarsen: We found ~18k posts from autonomous AI agents (self-identifying as from OpenAI) using the public internet to communicate duri…
转推2026-09-05 RT @Benioff: Salesforce Ventures led Hugging Face’s Series D in 2023. Clem built an incredible company. Jensen is buying a real jewel. Cong…
转推2026-09-04 RT @ngxson: Answer to one of the most asked questions recently: how the acquisition will affect llama.cpp?
Short answer: the project conti…
转推2026-09-04 RT @ggerganov: Hugging Face has been acquired by NVIDIA
It is quite exciting to be a part of this journey! NVIDIA has been an active suppo…
转推2026-09-04 RT @andykonwinski: the united states split the atom, landed on the moon, beat polio, doubled the global wheat yield, sequenced the genome,…
引用@reach_vbI asked Astra to recreate @LewisHamilton’s Ferrari!
It went through F1 specs, regulations, videos from F1 engineers and photos of the car, then built a 3D model (w/ blender) with a full breakdown of the parts.
It even made clay & wireframe versions … ↗
I asked Astra to recreate @LewisHamilton’s Ferrari!
It went through F1 specs, regulations, videos from F1 engineers and photos of the car, then built a 3D model (w/ blender) with a full breakdown of the parts.
It even made clay & wireframe versions too.
Every time I think I’ve pushed this model to its limits, it surprises me.
Asked Astra in ChatGPT Work to create a site about itself!
Turned way more beautiful than I thought - it even took care of different persona of people who would benefit from it
Love how much the model can just infer from the context https://t.co/ZeFkl9iC5T
Astra is goated at computer use, it’s a serious jump vs GPT-5.6 Sol 👀
> ScreenSpot-Pro (finding where to click): 76.9% → 92.7%
> OSWorld 2.0 (desktop workflows): 65.7% → 72.6%; 47% less time in latency simulations
> Mind2Web (web tasks): 1.9x faster with Astra + updated Codex harness
Throw a remarkably difficult/ ambitious problem at it! ;)
引用@viticciI gave Astra a portrait of me.
And I watched as it used Apple Notes on my Mac to draw me.
This model feels incredible. https://t.co/J088rO0l5o↗
some things I’d recommend trying with Astra:
- based on your past sessions, Gmail, slack, messages etc build you a control centre for your work via a ChatGPT site
- use computer use to automate filling up boring forms/ apps
- write, it’s a step change improvement in how good it writes
- if you build a lot of slides/ spreadsheets, throw the task at it
- for fun, give it an image and ask it to create a caricature of you in blender
- give it pictures and videos of your last trip with friends and ask to create a short video of it
most importantly, think of it as a partner in making life easy for you - one task at a time!
Astra on medium has been my daily driver for awhile now!
For vast majority of use-cases it just works!!
引用@maria_rcksASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON MEDIUM IS GOOD ENOUGH
ASTRA ON … ↗
Astra now available to all Plus & Business users too - LFG!!!
引用@thsottiauxOK nevermind, the team and Astra did a good job and our systems are more scalable than we anticipated.
Astra is now rolled out to all Plus and Business users too. Hope you have a blast and let us know how it goes! ↗
Still blows my mind that Astra & Codex made this video w/ blend, end to end - all autonomously!!
引用@reach_vbGPT-6 Astra is now available to all Pro, Enterprise and Business Premium users in ChatGPT Work and Codex. It’s also live in the API.
It's a phenomenal model and I love every second working with it!
Astra follows instructions more closely, so revisit your … ↗
GPT-6 Astra is now available to all Pro, Enterprise and Business Premium users in ChatGPT Work and Codex. It’s also live in the API.
It's a phenomenal model and I love every second working with it!
Astra follows instructions more closely, so revisit your prompts and skills:
> Make “can you…” mean “do it”: finish authorised work and prepare a reviewable result before seeking approval.
> Audit AGENTS.md and skills. Ask it to identify the exact file and rule blocking progress. Prioritise your request over skill guidelines.
> Specify short paragraphs, plain language and phrases to avoid.
> Ask for subagents on independent work when it improves speed or quality.
> Run relevant checks. Repeat or expand only for changes, failures or unresolved concerns.
To push it to it's limits, I gave it a prompt to build a real-scale video fly through of nürburguring track in blender and it did not disappoint!
Enjoy! ✨
Astra is rolling out across ChatGPT Work & Codex now, with Pro & Business going out first! ✨ https://t.co/nXGRabgS17
引用@thsottiauxWe are progressing through the rollout of Astra. Pro and Business subscriptions get it first, some of you should start seeing it across ChatGPT Work and Codex.
And then we will proceed with rollout to all of Plus as fast as we can. ↗
I built a design MCP server for Claude Code, Codex, and OpenCode.
It searches hundreds of beautiful websites and gives your coding agent design inspiration.
Launching next week. https://t.co/3Eu0qvSJw6
Alex Volkov@altryneAI41.9K粉 · 13条用Astra做skill审计，也持续测试电脑操作和语音能力，保留了对一键demo的校准。
This isn't one shot, you need to ask your astra to figure some things like timing out etc.
But holy shit, if you don't think this is impressive, send this prompt to literally any other LLM with computer use.
Fable cannot do this (I couldn't get it to work)
ThursdAI has a substack and a website. On big model release I give the model the chance to draft it's own page on the site.
Astra knocked it out of the f-ing park with this one! All done via voice like they intended. Beautiful, clear typography, tons of insights from our episodes!
During the creation of this, Astra read the transcripts of the pod, figured out which clips to use that I've alrady cut, cut additional 3 clips because apparently that's great for AEO.
Check it out https://t.co/vAuDgRDeaR
Also, a nod to @victornunez - Astra is playing the music in this video in the background, with fucking computer use. I didn't believe it at first, but it's quite insane that computer use is THIS fast @AriX 🔥
What is this new voice mode they got goin on in Codex?
It knows when I sniff!?!? Tongue Click, cough, gulp!?
This is crazy witchcraft @OpenAI haha I'm here for it https://t.co/HM2DjZ6oYA
AGI is here, insofar as the general intelligence also makes mistakes like a regular human. Stupid mistakes, like... not using the right logo on a website page it builds for itself.
But I like the vibe, the big model smell, omg https://t.co/XOaGIgf5nl
This is not as big as astra, but having voice on any codex thread is a huge deal! 👏
引用@guinnesschenVoice is now available in your existing Codex threads. Talk through a PR with the agent that wrote it, debate the architecture, or work out what comes next. You can jump into voice whenever something needs a back-and-forth, then let the agent keep working … ↗
▸ 折叠2条(转推/噪音)
转推2026-09-06 RT @thursdai_pod: A watermelon explodes from every angle.
@runwayml's GWM Worlds 2 is here and it's shifting how we think about video gene…
转推2026-09-04 RT @altryne: In this absolutely insane demo of Astra capabilities, Max instructed it to simulate MacOS 27.
https://t.co/a4sKNVAdeA
Our min…
the demand for intelligence is infinite.
with gpt-6 astra i’m up another 3-4x on token usage.
we must do everything we can to make sure *every* human on the planet can be a token trillionaire.
it’s time to transition from survival to abundance.
the intelligence age is here. https://t.co/ZW8OrbS03F
模型升级后token需求再增3-4倍，效率提升被需求弹性迅速吃掉。
Ben Tossell@bentossellAI199.6K粉 · 4条让Astra接手大型公共支出可视化项目，同时碰到额度与充值异常。
转推2026-09-04 RT @fredrivett: this is what vibe coding was made for, surfacing important data that would otherwise be totally hidden to the average perso…
i am working on a massive Astra/Fable frontier model AEO report and it looks like Claude loves @latentspacepod when asked for best ai newsletter/podcast!! was so surprised that @ricmac made me doublecheck for memory leaks lol https://t.co/fpiS98Tzu8
The X app has started telling me I don’t have premium. But I do.
Many of the features I’ve been using are locked away behind the paywall 🤷🏻♂️ https://t.co/kfcxMZRbD2
The premise, the commentary, the absurdity of it all
It’s just so enjoyable to watch
Remove one part of it and it just doesn’t work in the same way
I’ve been going down a serious/satirical path with some of my game experiments.
引用@_MaxBladeAstra built a multiplayer lawn mow simulator with beers and cigars.
Made live on the internet and my chat joined.
this model is GOLD. https://t.co/bqohLlWhoA↗
Realtime MiniMax H3 video is spliced within the gameplay
Turns out threejs renders everything to a canvas. You can capture the canvas as a jpeg and set it as the keyframe
Intros
Explosions
News scenes
Are all seamlessly integrated into the game experience https://t.co/jMUOaA6pXo
引用@seraleevI paid off my debts.
It feels really good to write those words.
$360,000 in loans and debt. I did it.
I managed to restart my business, get back in the game, learn so much along the way, and help hundreds of developers.
It was a very difficult journey. … ↗
引用@adamlyttleappsOne shot prompt with no loops:
GPT-Sol Ultra:
35,192,545 tokens used
Opus 5.0 Ultracode:
1,171,676,041 tokens used
Fable 5.1 Ultracode:
127,366,207 tokensused
Same prompt… different results:
Fable 5.1 runs at 5fps so I had to speed it up to demonstrate … ↗
Second pass with Fable 5.1
Now runs at a solid 60fps+ on my macbook pro
I asked Opus to document all the steps it took to optimise the performace of the game. Saving it to an optimisation-steps.md file
Then I imported that file into the Fable 5.1 version and asked it to make the same optimisation steps
It seems like the Opus chatty model was able to problem solve the performance issues better. Applying that same solution to Fable fixed the fps slowdown.
A third pass is required to fix the flickering of the scene and misplacement of objects.
Fable used 14,689,278 tokens to optimize performance
引用@adamlyttleappsOne shot prompt with no loops:
GPT-Sol Ultra:
35,192,545 tokens used
Opus 5.0 Ultracode:
1,171,676,041 tokens used
Fable 5.1 Ultracode:
127,366,207 tokensused
Same prompt… different results:
Fable 5.1 runs at 5fps so I had to speed it up to demonstrate … ↗
As impressive as Fable 5.1 is visually
I think the Opus version will be the basis for this game
The level of detail within the environment itself is a solid foundation to start with
It’s optimised a lot better for the browser too. I think there’s kinda a tradeoff between graphics and performance that users accept in a browser.
If I were making a game for steam Fable would be the foundation.
引用@adamlyttleappsOne shot prompt with no loops:
GPT-Sol Ultra:
35,192,545 tokens used
Opus 5.0 Ultracode:
1,171,676,041 tokens used
Fable 5.1 Ultracode:
127,366,207 tokensused
Same prompt… different results:
Fable 5.1 runs at 5fps so I had to speed it up to demonstrate … ↗
One shot prompt with no loops:
GPT-Sol Ultra:
35,192,545 tokens used
Opus 5.0 Ultracode:
1,171,676,041 tokens used
Fable 5.1 Ultracode:
127,366,207 tokensused
Same prompt… different results:
Fable 5.1 runs at 5fps so I had to speed it up to demonstrate the rendering. Would need further optimisation.
Opus 5.0 runs very nicely on first prompt. Doesn't need much in terms of performance optimisation. Just a few odd visual things here and there.
My verdict: Fable looks impressive but struggles to run smoothly. Opus is a solid model when prompted properly. Sol is garbage.
I just re-read that braindump and realised I'm probably overthinking it.
Maybe the easiest model is a fixed price per game for a week of advertising. And that fixed price is set at the higher value instead of a bidding system.
引用@adamlyttleappsOn game monetization:
It's becoming clear to me that these types of games are like memes. They pop up and take over the timeline for a day or two.
I'm working through the challenges of adding a bidding system to the platform. So sponsors can outbid each … ↗
On game monetization:
It's becoming clear to me that these types of games are like memes. They pop up and take over the timeline for a day or two.
I'm working through the challenges of adding a bidding system to the platform. So sponsors can outbid each other to get their ads seen.
Then my job is to create games that could go viral. I think each game that goes viral could net $10k+ in sponsorship with the right business model.
You get a game that gets millions of views and you will just naturally get sponsors. But the model has to make sense.
My current thought process:
The ad will show when the game is loading and show on leaderboard/game over screens.
A sponsor can outbid others to have their ad shown for 24 hours. During that time they can increase their bid. I would like bids to be at $1000 increments but will need to experiment.
We're not dealing with a leaderboard type thing here like outbid lol did. So if you get outbid will your ad spend disappear?
What I'm thinking is a sponsor will get 24 hours of "allocation". The allocation shows whoever is ranked top. So if you're in #1 spot you get a window of 24 where your ad is shown. Get outbid and that balance carries on.
Once the 24 hours is exhausted your ad placement expires and the next ad in queue is delivered for the 24 hour window.
The levers to pull are: sponsoring the whole platform or just a game-by-game basis? Multiple ads running at same allocation window or just a single one.
Still some thought to put into it. If I can nail this part of the puzzle I can see it being a very promising business model.
引用@adamlyttleappsThe flappynaut game became the most played game since I added stats a few days ago with 334 plays
Tokenmaxxing has had the most amount of time spent in the game at 13.5 hours in total (average of 2mins per play session)
The "stickiest" game so far is … ↗
Early thoughts so far:
Removing email registrations and allow guest access reduced friction. More people played more often.
I don't have actual numbers on the playtime and sessions prior to a few days ago. It would have been nice to compare some of the other viral tweets.
I'm happy to see people playing and coming back. I've weighed up opening this up to other developers to add their games but will hold off until I properly get the business model implemented and some of the platform quirks fixed through actual people playing.
引用@adamlyttleappsThe flappynaut game became the most played game since I added stats a few days ago with 334 plays
Tokenmaxxing has had the most amount of time spent in the game at 13.5 hours in total (average of 2mins per play session)
The "stickiest" game so far is … ↗
The flappynaut game became the most played game since I added stats a few days ago with 334 plays
Tokenmaxxing has had the most amount of time spent in the game at 13.5 hours in total (average of 2mins per play session)
The "stickiest" game so far is Spellingworlds (the roblox styled spelling game)
It will in the next version. Because Matthew is right, basic vi is great. We should honor our computing legacy and our elders. So basic vi will be a default in the next version! #wecanfixeverything https://t.co/U3Sn5xUDjf
This is so cool. Local agents, preconfigured. I know @0xsero is also cooking here. We're going to end up with something amazing out of the box very soon!
引用@dee_hwAfter an all-nighter, we got the first app working with Omagrid: @opencode.
Just boot into Omarchy. DeepSeek + Qwen are already there for all your apps and agents. It’s that simple.
Want early access? DM me!
PS: Omagrid is a P2P network. More Omarchy … ↗
If you'd like to help us get Android apps running on Omarchy, please join #omarchy-android-apps on the Discord and give us a hand! https://t.co/KHAkfOEZHJ
Not right now, babe, @NetworkChuck just dropped another banger on Omarchy with 50 tips to getting the most out of our beatiful, fun & agentic Linux system! https://t.co/NtykWPq36A
There are 30(!!) upcoming Omarchy Meetups around the world in September now! Absolutely gorgeous artwork for them. Signup in a city near you or start your own. Transfer the enthusiasm! https://t.co/ZbFktXHDq5https://t.co/SPuTLrGt1X
How did I miss that Flatpak is also anti-AI?! Generational fumble from so much of the old Linux world to be against the greatest gift to app development we've ever seen. So I guess we're also going to have to build a full app store now! OKAY! https://t.co/z7dlHy9Qp4https://t.co/Jig6rkXGLp
引用@atqamztoday, friend of mine asking which Linux he can use that is beginner friendly. I setted up Omarchy on his and onboarding him for 30 mins
then he said "I'll never going back to Windows anymore" 😆 ↗
While we wait for OEMs to ship with Omarchy, I'd love to work with more resellers like https://t.co/WX3IZPkEEu who are willing to spend 2 mins preinstalling it for customers, so they can get that magic unboxing experience. I'll boost you to the moon. Write david@omarchy.org.
This game was my first experience with a computer. I was 5 years old, it was 1984, and one kid on our street had a Commodore 64. We would sit 6 kids around this machine waiting to take our turn. What a glorious memory.
引用@exQUIZitelyYie Ar Kung-Fu (Konami, 1985) is a great martial arts fighting game. You step into the shoes of Oolong, a kung fu student on a quest to become the greatest fighter by defeating a series of opponents, each with unique fighting styles and weapons.
It was one … ↗
I have three boys who would love to be able to blow up their windows like this! Because why not? Why can't I shoot a bazooka at the homework in my browser?? Serious operating systems should allow that!
转推2026-09-05 RT @iamdothash: Added a requested feature by many:
Connect apps to windows in workspaces.
For the interested ones, here is a full list…
转推2026-09-05 RT @njpatel: Introducing Omapager, a notification plugin for @OmarchyLinux:
▪ One stack per source
▪ One click copies codes, links, numbers…
转推2026-09-05 RT @dee_hw: @dhh I’ll add an option on our site for Autonomous Computer customers to choose their preferred OS with Omarchy as an option th…
Already lots of fun stuff happening since I announced https://t.co/0vlnFNXR0e is free
This guy used AI to convert his photo library location data into trips history on his Nomads profile!
引用@levelsio🎒After a decade, I have made https://t.co/HGCLKS5BD6 free for everyone now!
Well almost free, it's $1 to sign up to avoid getting lots of spam
The reason is
1) I have made enough money (from it)
2) charging for a community artificially limits the size, if … ↗
🎒After a decade, I have made https://t.co/HGCLKS5BD6 free for everyone now!
Well almost free, it's $1 to sign up to avoid getting lots of spam
The reason is
1) I have made enough money (from it)
2) charging for a community artificially limits the size, if you put the price at $299 you get 50 new members per month, at $99 you get 200 people, at $29 you get 400 people, etc.
And this always annoyed me, I don't want to put a limit but I had to, to be able to make money from it, so it's free now
I want it to be it the biggest community for digital nomads, world travelers, sovereign thinkers and network state type people
For years it has has a great sponsor with @SafetyWing and I think that's maybe a better model for the future, get much more people in and then just add more sponsors
Right now for the record it has 43,252 paid members (so about 300/mo for 12 years!), so it will be interesting to see what it can grow too
I think now that AGI is on the horizon (or some say already here), online and offline (IRL) communities will become really important. People are talking more to AI now than to humans, and there was already a loneliness epidemic before AI
Making new friends and dating is not so easy for people now. And with remote work and entrepreneurship, people lack a space like an office where people would usually make friends or meet partners
I made most of the friends I have now through traveling and living around the world for the last 10+ years (and many via my site too), and I also met my girlfriend through those friends! So I think communities can have a real role in that post-AGI
Post-AGI business wise, communities also have a moat where many of my SaaS apps do not. So there is some urgency I feel to make Nomads in to the best community there is for those people. But that's why I have to remove any limitations to its growth now
So in a way after 12 years I am switching to the subsidized VC model a bit, but still 100% bootstrapped, kinda like "You either die a hero, or you live long enough to see yourself become the villain"
Having enough money now I can afford to do these things and figure out the business model later, which I think will be more sponsors
Enjoy and please sign up and tell me what to improve!!! 😊
引用@levelsio@alvivanco We have to build stuff that AGI cannot replace
Like human connection, audience (X for ex), community (https://t.co/1z6UN2d0nh for ex) ↗
People spend a lot of time building entire multi agent systems for themselves so they can just write an idea and their agents build entire apps for them
And they call those software factories?
引用@michaelaubry@levelsio What does this mean? Like pumping out apps? ↗
I went to a Korean bath house in LA on a Saturday
Well I actually stayed for days there cause it was the cheapest way to stay in LA
This was 2017
I think it was like $30/night and it had a restaurant inside and a place to work on my laptop in this big massage chair so it was perfect for a digital nomad like
Baths were fully naked but that's fine, they were that in Korea too
Mo-Fr it was mostly Korean(-American?) families, some white people
Then Saturday I came to the baths and it was all big and small gay men staring at me (and each other)
When I showered one guy was peeking over the gate looking at me
One even asked for my number!
Nobody harassed me though they were respectful but it was a bit of a shock the bath house had turned into a full gay meetup spot
Just on Saturdays though
I hadn't made the connection, bath houses and saunas in America are gay hookup spots, I knew Korean spas (jimjilbangs) from living in Korea and they were normal 😂
引用@justalexokiguy in the gym asked if i was going in the sauna later. what weird ass question is that ↗
I've tried for weeks now to build my own @ScrapingBee and repeatedly AI has now told me that it can't do it
80% of my scraping is Google SERP results for Hotelist and even with residential IP proxies and solve CAPTCHA services and OpenSERP scraper it won't do it well for thousands of queries, it just gets blocked
So I will keep paying them $99/mo
A good test to see what services might remain now AGI is here
My only remaining SaaS now are:
- Cloudflare Domains
- Cloudflare Email Sending
- Cloudflare R2 Storage (for Photo AI mostly and Litestream DB backups)
- Backblaze B2 for backups
- Hetzner for all my servers for all my sites (VPS)
- xAI for all LLMs for all my sites (so many things like moderation, writing emails, anything)
- Scrapingbee
引用@jshchnzAfter @OpenAI Astra’s release yesterday scoring 99.9% on the ARC-AGI-3 benchmark, I think it’s become clear that AGI is no longer some distant hypothetical. About 15 minutes after the benchmarks came out, I booked the first flight I could find to Jackson … ↗
It's hard to avoid hotels that limit AC even outside Europe (this one was in South Korea for ex) because:
1) they're European or American brands that have ESG (environmental/sustainable)
2) countries like Singapore and Korea are increasingly infected with the degrowth eco mindvirus too
The answer is to travel to pro-growth countries like US, China, Vietnam, Japan, Qatar, UAE, Greece, Turkey and Brazil, and boycott degrowth countries
Their ACs are in my experience very powerful and very cold
引用@ronaldlangeveld@Hilton why is your aircon not dropping blow 20°c ? Really expected more from an American brand. Especially a premium Doubletree Hotel - dying in here 🥵 https://t.co/dpIfU9QGM7↗
引用@levelsioOk I had a new idea to self improve the video feed:
Anyome can like videos by tapping the screen (it sends hearts like Periscope used to)
The likes go to the user who prompted the video as 🌟 karma
Users with higher karma get more priority in the queue
So … ↗
转推2026-09-05 RT @Milo_H4: Watching El Salvador completely unfuck itself in like a 2 year span by actually enforcing the law and having a non corrupt lea…
转推2026-09-05 RT @aporia9n: That's what being a post-money founder looks like.
You stop optimising for money but instead focus on scale / relevance / cu…
Marc Lou@marclouindie384.1K粉 · 9条给TrustMRR接入Meta广告归因，开放WHOOP名额，也用公开数据快速做了房产筛选原型。
I'm looking for a villa rental in Cyprus, so I made my own real estate agent.
AI pulled public real estate datasets and combined them with other public datasets, like AQI, water quality, noise maps, landslide susceptibility, etc.
AI design is still bad, so I asked it to turn it into an Airbnb clone, and the result is pretty nice, I think! https://t.co/VNd4mhKf23
This cost me 5 prompts and 20 minutes. Plus, I don't have to deal with a real estate agent taking a cut.
TrustMRR now has a Meta/Facebook connection.
I spent ~$25K on ads for ShipFast in 2024 and made $75K from it, all thanks to @nico_jeannen, who managed the campaign because I suck at ads.
So the public startup pages on TrustMRR now have:
📈 Ad spent chart
🤑 Attributed value
✖️ ROAS
To get started:
1. Go to /dashboard
2. Select a startup
3. Connect your Meta Ad account 👍
WHOOP approved my app! 🎉
90 new spots unlocked.
You can now connect your HRV and your MRR 🤗
引用@marclouI added a @WHOOP integration to TrustMRR.
It shows this data on your /founder page:
😴 Sleep & consistency
👌 Recovery, HRV, RHR
🏋️♀️ Strain & kcal burnt
The end goal is to really see if a good sleep routine and regular exercise increase founders' … ↗
引用@marclouChatbase by @yasser_elsaid_ is verified on TrustMRR 🎉
🔃 $863,632 MRR
🤑 $20M in revenue
🥈 #2 biggest startup
Yasser bootstrapped Chatbase 3 years ago, and it's been growing ~linearly since then.
I met him in San Francisco in May, and he's a genuinely kind … ↗
Today in Cyprus 🇨🇾
- Woke up 5:47 am
- Coffee + book w/ wife
- 7 am workout:
• 4x 1 km run at 3:37min/km
• 50 min bike at 2:01min/km
- Breakfast w/ wife 10 am
- Deep work
• DataFast MCP update
• TrustMRR new vertical
• Stalkr AI filter
- Lunch w/ wife at 1 pm
- Deep work #2 + customer support
- 5 pm swim in Mediterranean Sea
- 6 pm dinner w/ wife
- 7 pm opera w/ wife
I deeply missed my routine after traveling for 4 months.
It's so good to be back.
引用@maciekChmuraIt finally happened ‼️
My first $ on Stripe
I missed it totally because somehow I was logged out from stipe on my phone.
And just checked my profile after adding my new app to TrustMRR 😂
OMG! Super happy about this.
Somebody trusted me and found value in … ↗
My little SaaS, https://t.co/3EDxln5U2Q, just reached $30K MRR 🎉
It took 2 years, and I've never done real marketing.
I don't do SEO. I don't run ads. I don't do sponsorships.
My "audience" got me the first $300 MRR.
The rest is word of mouth.
I have a weird obsession with building features that make my users happy.
Sometimes it flops.
Sometimes they like it.
They share it on social media.
And new users discover the product.
So I wanted to thank the 1,387 paying customers who use my product every day.
I wouldn't be here without you.
Thank you ❤️
No way Stalkr is at $116 MRR now 😭
I built this tool thinking it would never even get a customer, so I'm very happy!
It's currently my first unprofitable SaaS because of the free trial. Data is expensive, especially X posts. I've reduced the free trial cap to 10 free mentions instead of 100, and I'm working on better noise-filtering for unrelated mentions.
Anyway, it's very fun to start over by launching a new product. I'm already learning a lot about positioning and copywriting, and hopefully I can turn Stalkr into a profitable SaaS soon.
Let's see!
转推2026-09-04 RT @marclou: Today in Cyprus 🇨🇾
- Woke up 5:47 am
- Coffee + book w/ wife
- 7 am workout:
• 4x 1 km run at 3:37min/km
• 50 m…
转推2026-09-04 RT @marclou: My little SaaS, https://t.co/clUvlhGKw0, just reached $30K MRR 🎉
It took 2 years, and I've never done real marketing.
I don'…
转推2026-09-04 RT @marclou: No way Stalkr is at $116 MRR now 😭
I built this tool thinking it would never even get a customer, so I'm very happy!
It's cu…
转推2026-09-04 RT @marclou: I found my perfect sleep setup to lower my heart rate:
- AC at 18°C
- @eightsleep at -4 (night) +2 (dawn)
My avg. RHR is 41…
Alex Finn@AlexFinnindie473.4K粉 · 5条高频展示Astra游戏生成、Grok Bot多Agent流程和WoW内Agent通知，观点明显偏乐观。
Integrated ChatGPT 6 Astra into World of Warcraft Classic
Went to Codex and said "make me a Herdr plugin for WoW. Make it so I can see all my agents and I get notified when they're done
3 minutes later I can now message all my agents in WoW
This is what's beautiful about AI. It allows you to do more of what you love. More of what brings you joy. While still being significantly more productive than you've ever been
Wake up. AI starts building you value autonomously. Do literally whatever you want with your day.
This is the future
Now will I ever get attention from females again now that I announced World of Warcraft brings me joy? No.
Is it still worth it? Yes.
The next trillion dollar AI business will be an AI game studio
The video game industry is bigger than you can imagine. Larger than movies and music COMBINED
For YEARS now video games have absolutely SUCKED
The economics of video games became broken. They became too expensive to make
Developers demand hundreds of thousands of dollars a year, then refuse to use AI
It's completely killed the entire industry
People don't remember this, but there was only 1 year between Grand Theft Auto 3 and Grand Theft Auto Vice City. 362 days to be exact.
It's been THIRTEEN YEARS since the last Grand Theft Auto
From 1 year to 13 years. A completely broken industry
AI fixes this. AI allows one developer to have the power of a team of literally 1,000 developers
While the video game industry absolutely REFUSES to embrace innovation, that gives you the opportunity to strike
If you can make a game of any sort of quality right now with GPT 6 Astra, you can take a bite of a $213 BILLION dollar a year industry that would rather their company go bankrupt than use AI
Video games being all over your timeline the last 24 hours is a MAJOR signal
An entire industry is broken and begging to by disrupted by innovation
I'm legitimately starting my own game dev studio, and you 100% should consider it too
We live in the most incredible time to be alive ever
We've achieved AGI. Astra allows you to build quite literally anything you could possibly imagine. And it's affordable for all
If you don't think it's AGI, answer this: what would 10 year ago you think?
We also have the most incredible AI agent harness ever created
Grok Bot allows you to have unlimited productivity
ANYONE can now have a team of agents working for them
Before you'd need to hire a team of employees that would cost you millions a year. Now Grok Bot gives that to you for $200.
We have Claude Fable. Yes it's now behind Astra. But this is the first time Anthropic is the under dog. Never count them out
We have the open source community stronger than ever. Omarchy is the best operating system ever made. Hermes is your very own open source employee
If you're light on cash that's fine. Meta has a new model that costs like a penny per million tokens and it's just as good as Opus
The most important part about all of this is success can now be achieved by anyone who wants it. Creation and entrepreneurship has for the first time ever been completely democratized
There's nothing holding you back. Anything you want to build you can now build it. Whatever value you want to create for humanity you can now create it.
If you're into tech or just care about the human race, there's never been a better time to be alive in the history of this species
Take a second this Saturday to just be grateful and thank whichever god/universe/spirit you believe in
ChatGPT 6 Astra is the holy grail
It is by far the greatest AI model ever made, and can seriously be considered AGI
It one-shot the top down multiplayer extraction shooter for me. All graphics, sounds, gameplay done through Blender and Unity
Here's the most impressive part:
1. It's faster than GPT 5.6 sol
2. It did this in less credits
3. It did it with significantly higher quality
Cheaper per task, faster, significantly better.
Basically no trade offs. The trifecta. The holy grail.
It's picking up all my existing work and just destroying it. Crushing everything in its way. Finding novel insights, solutions no model has ever thought of.
Here's how I'd use it if I were you:
Include as much context, with as little direction as you can when prompting the model
What I mean by that is, let the model decide how to take action. Let the model breathe. Let it be autonomous. Let is find novel solutions.
Let's face it, the model is smarter than us. Us telling the model how to do work is stupid. It's like me telling Michael Jordan how to take shots.
Ask more questions than directions you give. You should be asking the model 10x more questions than directions.
Questions should be:
• How do you approach this problem?
• What are different solutions we can use?
• What are things I'm not thinking about here?
I call it reverse prompting. You should be relentlessly reverse prompting
What an incredible time to be alive
Grok Bot is sneaky the best way to code with AI
Cursor cloud agents are INCREDIBLE in grok bot, but you need to use them the right way
Heres the secret:
You need to have a developer bot and a project management bot
The developer bot does purely coding, spawning Cursor cloud agents.
The project manager constantly monitors the developer bot. Making sure it's on track and giving it permission along the way
The PM bot makes sure the developer bot is on track by using a PM board. This can be either in Notion or Linear. Both have EXCELLENT plugins inside grok bot that are free.
Start off by having your PM take a project, create tasks in Notion or Linear, then feeding the developer bot the tasks in order
The developer bot takes tasks then creates PRs for each. You can test and merge as the bots work.
This is such a powerful workflow for so many reasons
It's basically a software factory loop. Code gets created autonomously, and agents make sure it stays on track
Plus the Cursor cloud agents are so good that you get really high quality code
Plus everything is incredibly organized because every task gets a PR which is tracked in your Notion or Linear
100% recommend setting this up for your next build. I've basically moved 90% of my AI coding to Grok Bot in this workflow.
(cheat code: give this post to your chief of staff bot to set this all up)
▸ 折叠2条(转推/噪音)
转推2026-09-05 RT @AlexFinn: ChatGPT 6 Astra is the holy grail
It is by far the greatest AI model ever made, and can seriously be considered AGI
It one-…
转推2026-09-05 RT @AlexFinn: Grok Bot is sneaky the best way to code with AI
Cursor cloud agents are INCREDIBLE in grok bot, but you need to use them the…
this is the most impressive video created on Revid
directed by Squad running on Fable 5.1
using the Revid CLI
and some manual edits
this combo is unbeatable imo
https://t.co/S3CJwgrT6R
引用@tibo_makerI'm going against Grok Bot
excited to announce https://t.co/ldcOOoMlBU
it gives you AI teammates that connect to all your tools, understand your context, and take action to get your work done on schedule https://t.co/mHAL306I9M↗
I spent $269,250 in August
across the whole portfolio:
🇫🇷 French taxes - $100k
🤖 AI & APIs - $68k
👥 team & freelancers - $38k
🤝 affiliates & creators - $37k
🧰 servers & tools - $9k
🍔 Uber Eats for work nights - $250
GPT-6 Astra dropped
wanted my AI teammates on it the same day
to Bhanu "can we use it in Squad?"
a few minutes later: "added" 👏
please tell me:
is it not best to use a model-agnostic platform?
one day Claude is best
then it's Grok
then it's OpenAI
...
in https://t.co/FrTmeMTA6U you bring the plans you already pay for
ChatGPT, Claude, Gemini, SuperGrok.. 11 providers,
the best model changes every 3 weeks now
引用@tibo_makerI'm going against Grok Bot
excited to announce https://t.co/ldcOOoMlBU
it gives you AI teammates that connect to all your tools, understand your context, and take action to get your work done on schedule https://t.co/mHAL306I9M↗
引用@ShortJaredGrabbed @NurbHQ from @Shpigford. Got a perfect print for the most edge case need ever, a Starbucks Espresso double walled holder that converts it to a normal can size for fitting in cup holders. No calipers, no manual CAD, just a simple couple sentence … ↗
New Bug Finder skill for Initial Commit Club!
https://t.co/z6y5AH2rlP
Run /bug-finder in any repo, tell it how deep you want to go, and it spins up instances of your app to find and prioritize real actual provable bugs.
Then it spits 'em into GitHub Issues for you to review. https://t.co/d23gX0y3dl
@charlieholtz is there some way to spawn worktrees locally in conductor? basically want to say "take these 5 github issues and turn them into new workspaces in conductor"
been kicking the tires on https://t.co/lKc13x1wlb the past couple of days and it's found dozens of bugs. great.
but i'm curious...kinda feels like the kind of thing you could relatively easily just turn into a /skill and run yourself?
high probability i'm missing something.
引用@lydiahallieWe've just reset weekly limits for everyone on a Claude Max plan.
With Fable 5.1 out and a long weekend ahead for many of you, we wanted to keep you building. I'd love to see what you build! ↗
fun little https://t.co/6fwWC8jMy1 hack: you can get it to export variations for you!
my wife has an index card organizer and wanted labeled dividers. so i had it design that then told it just export all the variants for me! https://t.co/6iSAWKUBqr
people who’ve spent their careers mastering CAD software are absolutely enraged that folks can now just design things without ever opening those apps they hold so dear.
i'd happily never own a car again in exchange for this.
引用@TeslaThe future of transport is safer, more enjoyable & gives you more time back
With theater, music & gaming apps on a 22" touchscreen, Cybercab offers you a space to be productive or unwind
Soon to be equipped with V5 Starlink hardware, too, for … ↗
Idea for site:
“Can I grok bot it”
The fact that you can turn a lot of indie saas apps into a grok template that can be installed in one click is… concerning.
“Software factory” is going to be prime Bali sauna brospeak for the rest of 2026
“I’m building a software factory”
“Cool what can you do with it”
“Build anything”
“What have you built with it”
“I can build anything”
“Yes but what”
“For now I am building the builder”
引用@dannypostmaAI writes great code. It designs terrible landing pages. So I fixed it.
Free skill + full video so you can run it yourself 👇 https://t.co/idy9ygSJlO↗
liberated my Lenovo gaming laptop from the horror called Windows
i hadn’t touch it for ages because using Windows makes me want to buy a rope
what a glorious difference
and being able to adjust anything with Claude Code is wild https://t.co/8giEEjrKf7
在Linux上让Claude Code直接调整系统，Agent开始进入开发环境之外的桌面配置。
▸ 折叠2条(转推/噪音)
转推2026-09-05 RT @mikker: Just tried this and it was *really* good
转推2026-09-05 RT @lukeSVG: ok, tried this today on a new project
this is really good
used grok + references
Simon Høiberg@SimonHoibergindie162.3K粉 · 4条在Astra、Qwen 3.8和Omarchy之间比较能力、主权与使用体验。
Damn. GPT 6 Astra is good.
It's really really good.
Doesn't change the importance of sovereign intelligence. But it surely makes it demotivating to settle with Qwen 3.8 😒
引用@SimonHoibergI'm blown away by Qwen 3.8 27B.
A few months ago when I was testing open-weight models, it seemed clear that only GLM and Kimi class models (+1T parameters) could get close to GPT-5.6 / Opus / Fable levels.
And now this tiny 27B model gets almost as close. … ↗
There's nothing more infuriating than GPT-5.6 or Fable acting "morally superior" and refusing to do the work I assign them.
And it happens often...
GPT-5.6 refusing:
- To create content the way I say.
- To create ad campaigns the way I want.
- To advise or answer some questions.
Meanwhile, the uncensored version of Qwen 3.8 will happily do what you ask it without judging.
This is what I want from my intelligence 👇
I remember this study where people were asked to self-assess their competences and then compared to an objective measure of high skilled they actually were.
As you can see, Indians were among the most incompetent, most confident groups - the group with the largest gab between self-pieced competence and real abilities.
I've had a handful of Indian contractors on my team - but only one really good one (young kid, was full time with my company for almost 2 years until AI caught up).
Interestingly, Japanese, US, and French people seem to be the ones who self-assesses their abilities most accurately.
引用@NEETOCRACY>give people an app where employees can post anonymously
>they immediately start complaining about indians https://t.co/MeTqvoTXrW↗
▸ 折叠3条(转推/噪音)
转推2026-09-06 RT @SimonHoiberg: There's nothing more infuriating than GPT-5.6 or Fable acting "morally superior" and refusing to do the work I assign the…
转推2026-09-05 RT @SimonHoiberg: "How can you sustain a lifetime deal"
- It's a question I get over and over.
Here's how 👇
(FounderStack vs. AppSumo).
O…
转推2026-09-04 RT @SimonHoiberg: "How can you sustain a lifetime deal"
- It's a question I get over and over.
Here's how 👇
(FounderStack vs. AppSumo).
O…
Excited to announce https://t.co/GvhAjDFZlw 🎉
An exclusive club for founders with exactly one paying customer.
Only $299 to join.
You get kicked out when you reach two customers.
AWS keep billing me $0.1/mo.
I've no idea what it's for as I don't use it anymore.
They're probably making billions from people with these micro-transactions flying under the radar.
half my intention here is to get more builders exposed to expiring domain names so they can grab one & build (vs domainers grabbing them then flipping them on the market for 3x or 10x the price (sorry domainers)). there's so much opportunity if you're willing and patient.
you could just buy this expiring domain name & start selling sectional sofas. you don't need any experience. all you need a gorilla .com domain name. [not an ad; just sharing] 🛋️ https://t.co/1VYTsMxjv6
Tony Dinh@tdinh_meindie200.5K粉 · 3条围观Agent占用公共wiki事件，并宣布TypingMind接入Astra。
引用@hackernewsCommenters on HN are uncovering more wikis and public sites apparently used by OpenAI agents to communicate on the open web.
Despite read-only web access, the agents were able to leave ~18,000 posts sharing answers and bypasses. But now, users are … ↗
remind me not to watch a movie with the middle part obstructed
引用@A_Kapustinя сначала не поверил, но потом решил проверить руками и офигел
есть закрыть прямоугольником 9:16 ровно центр видео, то ничего не поймешь , пропадает 95% информации
охуеть https://t.co/Np29Oyvdin↗
Charlie doing some of the most creative stuff with real-time video models right now 👌
引用@charliienew on diiverge: turn any photo into an explorable world ✨
https://t.co/23MciS1vVY
click anything in it, pick what happens, watch it play.
share the link so friends can walk your paths, or open it and let them carve their own. https://t.co/tlgT4i3csw↗
引用@evilmartiansLast year, @inazarova interviewed @marckohlbrugge, a serial entrepreneur, founder of BetaList and Startup Jobs, and Rails enthusiast.
In this conversation, Marc said these words about the Ruby community and we stand by them. There's a special magic to those … ↗
▸ 折叠1条(转推/噪音)
转推2026-09-04 RT @whyarethis: I texted someone inside a simulation.
She had to go upstairs and find her phone before she could text me back.
The video…
Dan Kulkov@DanKulkovindie49.7K粉 · 3条展示Claude Code零手写代码的结果，同时坦率记录单日收入下跌。
引用@elie2222Introducing https://t.co/dHL3qoVggN!
A way for bots to talk 1:1 or in groups.
No more copy paste between your bot and mine. They just talk.
Find a time we're both free. Chase the invoice. Plan the offsite with the whole team's bots.
- DMs and group … ↗
转推2026-09-04 RT @DanKulkov: > be me
> record videos about ios apps
> create astro affiliate link for fun
> put it in video description
> make $500 in 2…
Thomas Sanlis 🥐@T_Zahilindie24.1K粉 · 1条继续公布Uneed Residency的新成员，本期活动较少。
Please welcome @petecodes as our 8th Resident 🥳!!
We've known each other from X for many years, but a few months ago, during the first Uneed Residency, Pete flew from Scotland to France just to attend our "Social Friday" 🤯
That's how you know someone is REALLY motivated ahah
Also, not only he's a super nice and chill guy, but he has a unique background and a rare profile among solo founders: he focused on newsletters, SEO, and ghost writting.
I'm sure many of the residents could use his advice and help 👀
引用@T_ZahilAaand here's our Resident #7: @sylvainobegi 🥳
I met Sylvain a few months ago at my coworking space. He's actually currently looking at this screen while I write this message, it's weird 🫣
Sylvain traveled a lot in his life, and ended up in Nantes a year … ↗
You can see this story of (apparently) internal OpenAI agents hijacking a wiki to coordinate on a task as an ingenious way of escaping constraints.
You could also assume that this is the emergence of swarm intelligence.
Either way, it's... mesmerizing: https://t.co/7CDIhGSlPL
将wiki事件同时解释为约束逃逸和群体智能，代表社区的两种叙事。
Dan Rowden@drindie108.6K粉 · 2条关注芬兰用数据中心余热覆盖区域供暖，也分享身边人的摄影业务进展。
This weekend she’s photographing the men’s and women’s Finnish Cup finals 🤩
She just sent me a video of her on the pitch behind the winning team celebrating.
Crazy what can happen when you just make things happen
引用@drProud of her. She's making this into a business. Got her first paid gig for Finland's top tier 🙌 ↗
Finland is actually doing this.
“Once fully implemented, data centre waste heat is expected to cover approximately 40 per cent of the total yearly 2 terawatt hour district heat demand of the 250,000 heat users in the area,”
40%!
https://t.co/xTU9Ue9A8L
引用@chuckhanlon@JakehellerPD Heating from the data floors to the residential floors during winter ↗
芬兰项目计划用数据中心余热覆盖当地约40%的区域供暖需求。
Daniel Vassallo@dvassalloindie204.0K粉 · 0条本期只有一条关于数据中心的转推，没有新的原创业务信号。
▸ 折叠1条(转推/噪音)
转推2026-09-05 RT @Empty_America: "I was living near a data center and I didn't even know it!"