Feed
What I'm posting on X.
Posts, replies, reposts, and quotes from @jasucho. Archived at build time, newest first.
AUGUST 2026
We're making Git hosting more reliable, performant, and scalable.
This post traces 20 years of Git infrastructure and explains how that history led us to design and operate our Git storage, Origin, as if it were a database.
cursor.com/blog/git-at-any-scale
This post traces 20 years of Git infrastructure and explains how that history led us to design and operate our Git storage, Origin, as if it were a database.
cursor.com/blog/git-at-any-scale
@samsaffron I don't think many humans are going to be reading or writing code in 5 years, so I don't really think it matters.
It turns out you can run a full browser implementation in a Worker. No container needed.
I didn't believe it myself when they first told me. Absolutely crazy project.
blog.cloudflare.com/kitesurf
I didn't believe it myself when they first told me. Absolutely crazy project.
blog.cloudflare.com/kitesurf
JULY 2026
Kimi K3 is now available on Ollama’s cloud.
To use it with Claude Code, run:
ollama launch claude --model kimi-k3:cloud
Currently Kimi K3 requires a Pro or Max subscription, and consumes extra usage credits. We’re quickly working on adding capacity to expand access.
To use it with Claude Code, run:
ollama launch claude --model kimi-k3:cloud
Currently Kimi K3 requires a Pro or Max subscription, and consumes extra usage credits. We’re quickly working on adding capacity to expand access.
@Polymarket I really like @AnthropicAI models and products in general but stuff like this just genuinely makes me want to support @OpenAI and open models from Chinese labs
This is really cool, there are some awesome designs in here. Hope a bolder design wins 💪
Claude Code on desktop now works with the iOS simulator.
Build and run your iOS app, and the simulator opens in a panel right next to your conversation. Available today in public beta.
Build and run your iOS app, and the simulator opens in a panel right next to your conversation. Available today in public beta.
Let me show you how to run Kimi-K3 locally, right from your own house
i was really dissapointed with 5.6 at first because out of the box its burning tokens like crazy in codex but now i switched to using it in @pidotdev and now i get it. its cheap and it just gets stuff done. just need the right harness to let the model shine.
Morning. The last 48 hours of Codex and ChatGPT Work have been intense! Three important updates:
- Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans
- Rolling out changes that will make GPT 5.6 Sol more efficient across the board and that will be reflected in less usage being used so that it can take you further. Exact impact to be quantified and shared
- We hit 6M active users, and are landing a usage reset in the next hour
Go do things
- Temporarily removing the 5 hour usage limit restriction for all Plus, Business and Pro plans
- Rolling out changes that will make GPT 5.6 Sol more efficient across the board and that will be reflected in less usage being used so that it can take you further. Exact impact to be quantified and shared
- We hit 6M active users, and are landing a usage reset in the next hour
Go do things
I love Sol but you definitely should keep it on a leash. It’s a bit too willing to do anything to complete its work
SpaceXAi overtaking Google to become the third place lab is such a hilarious and sad unforced error from Google
Cannot wait to find out in 5-10 years what actually went so horribly wrong
My current ranking:
1) OpenAI & Anthropic (tie)
3) SpaceXAi
4) zAI
5) Google
Cannot wait to find out in 5-10 years what actually went so horribly wrong
My current ranking:
1) OpenAI & Anthropic (tie)
3) SpaceXAi
4) zAI
5) Google
GPT-5.6 Sol, along with Terra and Luna, will launch publicly this Thursday.
We’re expanding preview access globally now.
We’re expanding preview access globally now.
Cloudflare is on a roll. They keep making cool things. At this point i am fully Cloudflare native.
@Pratham_JOD @theo @jullerino Compared to previous models and especially if you have the golden gate Beta running with the new device hub fable is far better at testing apps in the simulator. Never gets old watching it click and scroll like a maniac.
@asklumo 2.0 Max is really good. Easy daily driver ready for every day AI questions.
Definitely better daily AI answers than Mistrals app currently if you want to euromaxx.
Definitely better daily AI answers than Mistrals app currently if you want to euromaxx.
@SemiconductorsX It sounds cool but i would be curious to read why that would make sense and how they handle the challenges. Also are we getting an ocean region in aws now? 🛥️
Fable turned my remarkable into Tom Riddle's diary from Harry Potter.
My prompts fade, a LLM respond.
Magical!
My prompts fade, a LLM respond.
Magical!
I tweaked this a bit since I also wanted to add GLM 5.2 through ollama cloud with pi agent and so far the results are very good. I can work out a decent iteration even with my Pro plan.
Thats my current global CLAUDE.md:
a good engineer with kimi 2.7 will out perform an average engineer with fable any day of the week
@ParthJadhav8 Such a good movie and it was incredible at being eerie. Haven’t watched a movie that stressed me out so much in a looong time.
Interesting. Definitely need to play with it more. I feel like i am someone who is pretty deep into this stuff and it’s moments like these where i feel how fast the ai space is moving. Literally can’t keep pace with the way how we interact with the model changes.
@ollama also if someone is interested i quickly vibed together a mac wrapper for lumo: github.com/svcho/Lumo-for-Mac.git
Need to improve it but works well as a first iteration. Of course not affiliated with @ProtonPrivacy
Need to improve it but works well as a first iteration. Of course not affiliated with @ProtonPrivacy
@asklumo from @ProtonPrivacy is actually good now for every day use. will daily it as my random question companion for now, excited to see how it fares longer term. great release!
shoutout to @ollama. glm 5.2 is super fast and fun to use. have found myself using it as primary LLM the past couple of days and its really capable. Open models are the future.
JUNE 2026
AI knows more about you than any other technology.
Most AIs mine your data to train their models and profile your behavior, which later gets sold to advertisers.
Lumo doesn’t. And it just got a massive upgrade.
Here’s Lumo 2.0, the world’s most-private AI ⬇️
🧵 1/6
Most AIs mine your data to train their models and profile your behavior, which later gets sold to advertisers.
Lumo doesn’t. And it just got a massive upgrade.
Here’s Lumo 2.0, the world’s most-private AI ⬇️
🧵 1/6
@cursor_ai Update: it works now in Austria 🇦🇹 👏
@badlogicgames If you want to euromaxx i would go for fairphone (which is most likely a downgrade from the 9a) otherwise i would stay pixel with android.
@badlogicgames This is the funniest article i have read in a while. I love Austrian politics (sarcastically of course)
@OfficialLoganK @MeechYourGoals @GoogleAIStudio An official Google Play Store Developer CLI that could give an agent the tool to autonomously publish a new app version would be awesome in general. Play store submissions are beyond terrible.
understanding every PM ive ever interacted with right now
what is bro cooking
we need to ship
what is bro cooking
we need to ship
@deepseek_ai is the best. Thankful that there are labs who operate this openly in times where access to ai models gets more restricted.
Good new first: Sol is a smart, efficient, and a significant step forward. It is the same price as GPT-5.5. Also launching in the GPT-5.6 family is Terra, with 5.5-level performance at half the price.
Bad news: at the request of the US government, it is launching today in limited preview instead of the open access launch we were planning on. We are working with the government to get to general availability as fast as we can.
I think it is quite reasonable to roll out models--especially as they reach significant new levels of capability--in this way. It fits with our long-held strategy of iterative deployment. But this isn't quite the process that we think is optimal.
Now we will with the government to attempt to get to a transparent, reliable process for early access, and to ensure that as long as our safeguards work as intended we can release widely. We want to be a reliable, dependable partner that works with all stakeholders, and we also want to live by our mission of benefiting all of humanity. I believe the government shares most of our goals, and that they are overall doing a good job in a very difficult situation.
We will work as quickly as we can to get this model in your hands and we hope you will love it.
Bad news: at the request of the US government, it is launching today in limited preview instead of the open access launch we were planning on. We are working with the government to get to general availability as fast as we can.
I think it is quite reasonable to roll out models--especially as they reach significant new levels of capability--in this way. It fits with our long-held strategy of iterative deployment. But this isn't quite the process that we think is optimal.
Now we will with the government to attempt to get to a transparent, reliable process for early access, and to ensure that as long as our safeguards work as intended we can release widely. We want to be a reliable, dependable partner that works with all stakeholders, and we also want to live by our mission of benefiting all of humanity. I believe the government shares most of our goals, and that they are overall doing a good job in a very difficult situation.
We will work as quickly as we can to get this model in your hands and we hope you will love it.
It is very easy to be pessimistic and right.
Most ideas are bad. Most markets are too early, too small, too crowded, too hard. You can build an entire worldview around seeing the flaw first, and be rewarded for it over and over again
But the strange thing about startups is that the only outcomes that matter come from the places where someone was optimistic and right
I used to underestimate how much this matters. But there is a whole world of difference between reacting to the world as it is, and having enough understanding of the world as it is to still stay open to what it could become
Long optimism
Most ideas are bad. Most markets are too early, too small, too crowded, too hard. You can build an entire worldview around seeing the flaw first, and be rewarded for it over and over again
But the strange thing about startups is that the only outcomes that matter come from the places where someone was optimistic and right
I used to underestimate how much this matters. But there is a whole world of difference between reacting to the world as it is, and having enough understanding of the world as it is to still stay open to what it could become
Long optimism
@mitsuhiko Derstandard is borderline unreadable nowadays. Their print version is still pretty good though.
How have i never heard of this 👀
How likely is it that /goal add Docker compatible composer support works for this
How likely is it that /goal add Docker compatible composer support works for this
his take is a lot more nuanced than the tweet makes it out to be... Torvalds is not a luddite.
Needless to say i will have to downgrade even though Golden Gate is a great update even in beta 1 🫡
So far so normal but the issue is that the stable version of Xcode does not run on Golden Gate anymore meaning that when you switch your only device to the beta you are unable to ship any updates to your Apple apps.
Do not update to macOS Golden Gate beta if you are an apple developer!
The Midjourney medical thing is genuinely strange and I kind of love it.
The plan is a spa.
Hot tubs, saunas, cold plunges, open 24/7, first location in San Francisco in 2027.
You step into a shallow pool of water, sink slowly through a ring of half a million tiny ultrasonic sensors, and in about 60 seconds you walk out with a 3D map of your insides down to a fraction of a millimeter. No magnets, no radiation, no contrast, just sound waves and warm water.
Compare that to how we do this now:
They say it's close to 100x faster than an MRI ("60 seconds"). For context, a normal MRI in the US averages around $1,300 and the scan alone can take over an hour inside a loud metal tube. A full-body scan from Prenuvo runs about $2,500 for roughly the same hour.
Midjourney wants to flip the whole feeling of it. Build a place you'd want to visit even if there were no scanner, then collect the health data as a side effect. I have no idea yet if the tech delivers what they claim. But the framing is smart.
The hardest problem in preventive health has always been getting people to actually show up, and a spa solves that better than a hospital ever will.
The plan is a spa.
Hot tubs, saunas, cold plunges, open 24/7, first location in San Francisco in 2027.
You step into a shallow pool of water, sink slowly through a ring of half a million tiny ultrasonic sensors, and in about 60 seconds you walk out with a 3D map of your insides down to a fraction of a millimeter. No magnets, no radiation, no contrast, just sound waves and warm water.
Compare that to how we do this now:
They say it's close to 100x faster than an MRI ("60 seconds"). For context, a normal MRI in the US averages around $1,300 and the scan alone can take over an hour inside a loud metal tube. A full-body scan from Prenuvo runs about $2,500 for roughly the same hour.
Midjourney wants to flip the whole feeling of it. Build a place you'd want to visit even if there were no scanner, then collect the health data as a side effect. I have no idea yet if the tech delivers what they claim. But the framing is smart.
The hardest problem in preventive health has always been getting people to actually show up, and a spa solves that better than a hospital ever will.
Really love the transparency. Rooting for a strong model this summer. We need more competition from Europe.
@GuillaumeLample This would be an amazing time for a Mistral comeback. Believing in the orange fat cat 🐈
i hope it's clear now why open source models are important
i've said before i can respect the position around safety but it's completely naive
even if you think you have superior morality and should control it someone will kick you out and take control
i've said before i can respect the position around safety but it's completely naive
even if you think you have superior morality and should control it someone will kick you out and take control
@ryanvogel This is why i love building for Apple. They provide great infrastructure as long as you stay Apple only which is a big catch
Two things I am excited about after the fable release:
1) OpenAIs answer and their teased app revamp
2) Whatever open Chinese models will come out trained using fable
1) OpenAIs answer and their teased app revamp
2) Whatever open Chinese models will come out trained using fable
This is a super exciting release - Claude Fable 5 is the same underlying model as Mythos but with added safeguards. The benchmarks are great and it's SOTA on everything by a margin but I'll add that *qualitatively* also, this is a major-version-bump-deserving step change forward (imo of the same order as Claude 4.5 was in November), peaking especially for long problem-solving sessions on very difficult problems. You can give it a lot more ambitious tasks than what you're used to, the model "gets it" and it will just go, and it's never felt this tempting to stop looking at the code at all (but don't do this in prod!). The model still has quirks that people will run into and the safeguards are configured to be a little too trigger happy for launch, which can hopefully be tuned over time.
I feel a lot of things changing as working software increasingly comes out on a tap. The Jevon's paradox kicks in and I feel my own demand for software growing substantially. You can ask for anything - explainers, visualizers, dashboards, bespoke single-use apps (e.g. a full wandb that is hyper-specific just for your project), you can 10X your test suite, auto-optimize code, run giant research projects with custom HTML for the results, anything! "Free your mind" (Matrix ref). Really looking forward to all the things people build!
I feel a lot of things changing as working software increasingly comes out on a tap. The Jevon's paradox kicks in and I feel my own demand for software growing substantially. You can ask for anything - explainers, visualizers, dashboards, bespoke single-use apps (e.g. a full wandb that is hyper-specific just for your project), you can 10X your test suite, auto-optimize code, run giant research projects with custom HTML for the results, anything! "Free your mind" (Matrix ref). Really looking forward to all the things people build!
Reposted media post
Apple made Siri AI the future of the iPhone, then locked European iPhones out of it. Here's what that actually tells us about privacy, the DMA, and where Europe is heading. 👇🔗
As a EU citizen, I'm really worried that the EU will disappear and be technology irrelevant.
It's one thing to not be able to produce the latest tech, but it's another thing to also cut yourself off from the latest tech because you have bureaucrats unable to understand a thing.
It's one thing to not be able to produce the latest tech, but it's another thing to also cut yourself off from the latest tech because you have bureaucrats unable to understand a thing.
Meet LM Studio's mobile app.
Your local models, now in your pocket.
Your local models, now in your pocket.
MAY 2026
Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors.
Available today at the same price.
Available today at the same price.
To simplify our Codex compute fleet management, we will be sunsetting GPT-5.2 and GPT-5.3-Codex in Codex on June 2nd when logged in with your ChatGPT account.
For free plans, GPT-5.5 will be the default frontier model to build and work with going forward.
These models will remain available on our API.
For free plans, GPT-5.5 will be the default frontier model to build and work with going forward.
These models will remain available on our API.
Whenever anyone complains about vibeslop I remember all the “production” codebases of billion dollar companies I’ve worked with.
It’s so rare to find a repo as clean as something like Pi.
Share your best oss codebase and I will poke around in there. (:
It’s so rare to find a repo as clean as something like Pi.
Share your best oss codebase and I will poke around in there. (:
Gemini Omni Flash is rolling out starting today.
Here’s where you can find it:
🔹 Today: Google AI Plus, Pro and Ultra subscribers globally in the @GeminiApp and @FlowbyGoogle .
🔹Rolling out starting this week, for no cost: @YouTube Shorts and the YouTube Create app.
🔹Coming weeks: developers and enterprise customers via APIs.
#GoogleIO
Here’s where you can find it:
🔹 Today: Google AI Plus, Pro and Ultra subscribers globally in the @GeminiApp and @FlowbyGoogle .
🔹Rolling out starting this week, for no cost: @YouTube Shorts and the YouTube Create app.
🔹Coming weeks: developers and enterprise customers via APIs.
#GoogleIO
Both are genuinely capable. Neither is perfect. And honestly, half the joy is just finding excuses to resurrect a Raspberry Pi that was collecting dust 🛠️
That being said OpenClaw’s Ollama integration handles model serving better under the hood. More stable once it gets going, even if the experience getting there is bumpier.
That being said OpenClaw’s Ollama integration handles model serving better under the hood. More stable once it gets going, even if the experience getting there is bumpier.
OpenClaw is rougher around the edges. Slower, needs more tuning, and I still haven’t got it feeling as smooth as Hermes. Time to first token is long. Sometimes I genuinely couldn’t tell if my Pi was doing anything at all 😅
Been testing Hermes Agent and OpenClaw on a Raspberry Pi 4 for the past couple of weeks. Both running DeepSeek-V4-Flash via Ollama Cloud. Here’s what I found 🧵
Been testing Hermes Agent and OpenClaw on a Raspberry Pi 4 for the past couple of weeks. Both running DeepSeek-V4-Flash via Ollama Cloud. Here’s what I found 🧵
My x feed is basically a mix of people hyped about the fully automated ai future where robots write every loc and people who think that we went too far and need to slow down in slopville. I like the two viewpoints. Makes me excited to see how everything is going to evolve in the next years.
Zen Kana 2026.5 (May Update) 🖌️
☀️ Daily recs so you never wonder what to study
✍️ Draw kana with animated stroke-order guides
🎯 Stats now filtered to your active selection
⚡ Firebase SPM migration, faster iOS builds
Less deciding, more learning 🇯🇵🧑🎓
Check it out: zenkana.app
☀️ Daily recs so you never wonder what to study
✍️ Draw kana with animated stroke-order guides
🎯 Stats now filtered to your active selection
⚡ Firebase SPM migration, faster iOS builds
Less deciding, more learning 🇯🇵🧑🎓
Check it out: zenkana.app
This weekend I’m begging you to look up from your laptops and remind yourselves that the world is a billion times bigger by size shape emotion feeling experiences than what you’re getting on here
This might teach you how to get better at B2B SaaS
This might teach you how to get better at B2B SaaS
@kimmonismus Unless they improve geminis toolcalling it’s still not going to be relevant for coding tasks.
That is smart. Googles biggest advantage is the vast amount of people using it for basically everything in their private life.
Also curious to see if Apple can pull something like this off since they are rumored to have partnered with Google on their Siri revamp.
Also curious to see if Apple can pull something like this off since they are rumored to have partnered with Google on their Siri revamp.
@svpino @FrameworkPuter Man if my MacBook ever dies it would be really tempting to try the framework 13 pro (with linux of course)
@badlogicgames I feel like the best laptop to buy is the one that you can/want to afford right now. There is always something better coming soon but the current macbooks are all great
After using hermes agent daily now with @deepseek_ai v4 flash i have high hopes that we will get very capable models that can do meaningful work on a reasonable machine in the near future 🐳
Also shoutout to @NousResearch. Hermes is awesome.
Also shoutout to @NousResearch. Hermes is awesome.
@ClaudeDevs First i was excited but raising the 5 hour limit without the weekly limit is basically bullying 😂
@Natebuildsai I think there are plenty of opportunities to find solutions for the humanitarian or social sector. They are left out most often but there you can have a profound impact with simple solutions.
Pangi Authenticator 2026.5 Release🔐
🔏 Recovery codes can be managed with your 2FA codes and optionally synced via iCloud (encrypted)
⌚️ Watch can clipboard-copy codes back to iPhone + Digital Crown navigation
🇩🇪 German localization landed, all strings properly keyed
The release where your secrets sleep soundly.
⬇️ Download: apps.apple.com/app/pangi-authenticator/id6760948720
🔏 Recovery codes can be managed with your 2FA codes and optionally synced via iCloud (encrypted)
⌚️ Watch can clipboard-copy codes back to iPhone + Digital Crown navigation
🇩🇪 German localization landed, all strings properly keyed
The release where your secrets sleep soundly.
⬇️ Download: apps.apple.com/app/pangi-authenticator/id6760948720
APRIL 2026
The whole vibe coding lifecycle is basically:
1) holy shit i can just build things
2) oh boy i am stuck, its over engineering random stuff and not doing what i want
3) ok i need to learn how to code to instruct my agent better
And i think thats a good thing. This might be the new pipeline to get people into coding 👨💻👩💻
1) holy shit i can just build things
2) oh boy i am stuck, its over engineering random stuff and not doing what i want
3) ok i need to learn how to code to instruct my agent better
And i think thats a good thing. This might be the new pipeline to get people into coding 👨💻👩💻
This 47-min interview with Boris Cherny (the creator of Claude Code) will teach you more about AI-native development than 6 months of trial and error.
Watch it, bookmark it, share it.
Your entire approach to building with Claude will shift.
Watch it, bookmark it, share it.
Your entire approach to building with Claude will shift.
fun codex hackathon, thanks for hosting and organizing @mrlesk @_Anastasia_Ana @DanielGri @dreym_dev @abbanbhan ❤️
LLM post-training used to mean fine-tuning to a downstream task
Robotics has been stuck in this setting, needing task-specific fine-tuning for best performance
π07 changes this: It works out of the box & outperforms fine-tuned specialists
Details: pi.website/pi07
Robotics has been stuck in this setting, needing task-specific fine-tuning for best performance
π07 changes this: It works out of the box & outperforms fine-tuned specialists
Details: pi.website/pi07
@joshwoodward @addyosmani I love that one of the side effects of agentic engineering is that there might be more native apps again. Great job!
Cloudflare Mesh is here. Ready to connect your devices, servers, and agents to a single private network 🔐
And with Workers VPC, your Workers, Agents, and Durable Objects running on Cloudflare can now reach your private MCPs, APIs, and databases directly
Oh, and it's 50 nodes + 50 users free on every account
And with Workers VPC, your Workers, Agents, and Durable Objects running on Cloudflare can now reach your private MCPs, APIs, and databases directly
Oh, and it's 50 nodes + 50 users free on every account
MARCH 2026
the future of saas in one interaction.
Is software turning into a liquid?
Is this what most software is going to be? Nameless, shapeless? Created (poured?) just in time, evaporating just after?
My current fascination in this week's Joy & Curiosity intro.
registerspill.thorstenball.com/p/joy-and-curiosit…
Is this what most software is going to be? Nameless, shapeless? Created (poured?) just in time, evaporating just after?
My current fascination in this week's Joy & Curiosity intro.
registerspill.thorstenball.com/p/joy-and-curiosit…
Just received my first mail to vote for a shareholder meeting through @Bitpanda_global. Really cool implementation and super easy. Will done 🙌
FEBRUARY 2026
Ich habe in 15 Minuten mit Codex ein Skill für Claude erstellt, um Echtzeitinformationen von der Wiener Linien-API abzugreifen und so einfach die nächste Abfahrt einer Haltestelle zu erhalten oder Routingsinfos von den @wienerlinien zu bekommen ✌️ enjoy
github.com/svcho/Wiener-Linien-Claude-Skill
github.com/svcho/Wiener-Linien-Claude-Skill
you can outsource your thinking
but you cannot outsource your understanding
but you cannot outsource your understanding