@antirez Do consumers get stuck with Opus 4.x class and GPT 5.x class for years while Mythos patches all software behind the scenes?
English
Jon Sherrard
1.3K posts
Jon Sherrard
@jshez
Building
Ciudad Autónoma de Buenos Aire Katılım Aralık 2022
1.2K Takip Edilen365 Takipçiler
Follow me on this: if Mythos benchmark numbers and model card info is correct (which I have little reasons to doubt), Anthropic can easily synthesize a better Opus, given they don't want to release Mythos. So, I bet we'll soon see Opus 4 at the level of GPT 5.4 or greater.
English
@waitbutwhy Analogous to British and American politicians and the 4th estate. Media shames and criticise relentlessly -> Parties and electorate select for shamelessness, and insulation to criticism irrespective of their talent.
English
Just watched Louis Theroux’s doc about the manosphere, and I think you can draw a pretty straight line from “any display of male libido has been punished for the past decade” to “the only remaining visible masculine role models for many young men are actual misogynist dickbags.”
Camus@newstart_2024
The guy just landed a spacecraft on a comet — one of the most impressive scientific achievements in years. His reward? A public struggle session because his bowling shirt had scantily clad women on it. Helen Andrews points out the quiet cost of institutional feminization: HR departments now hunt down any maverick personality and stamp it out. We’re losing innovators we’ll never even know about, all because someone focused on the shirt instead of the comet. This is how wokeness actually works. Have you seen real excellence get punished for something trivial like this?
English
So Mythos has to fix every bug on the internet, every package, every operating system, and every browser before it can be released.
Does that mean the general public is stuck on Opus 4.6 and GPT 5.4 capability LLMs for a few years?
English
English
English
if you're seeing this, make sure you turn thinking on
claws are actually pretty great in gpt-5.4, but they seem really stupid unless this gets turned on
Jeff Escalante@jescalan
Transition to gpt in openclaw not going so great this far 😂
English
@steipete @aleks_todo @DannyIO @danshipper Update to dev channel worked for me, but gpt-5.4 still regularly says "I've started that task", instead of starting->waiting for result-> replying
English
English
Yesterday I used @perplexity_ai Comet to plan some simple meals and snacks for the week, and then order it all at Carrefour via Rappi.
Amazing how the future is here and we're so blasé about it.
And as the old adage goes, it's the worst it'll ever be.
English
@bcherny @MattLeonard1616 Also reporting a not working but also the value does not match the value of my plan
English
Starting tomorrow at 12pm PT, Claude subscriptions will no longer cover usage on third-party tools like OpenClaw.
You can still use these tools with your Claude login via extra usage bundles (now available at a discount), or with a Claude API key.
English
Jon Sherrard retweetledi
That's us! 🌍
The Artemis II crew captured beautiful, high-resolution images of our home planet during their journey to the Moon. As @Astro_Christina put it: "You guys look great."
English
@badlogicgames Superpowers + @plannotator especially really helped me with this. I actually enjoy revising plans in depth now. Spend a lot more time on them.
English
just as with code, letting your agent create a plan mostly unsupervised leads to slop.
if the plan is so big that your attention span keels over, you have a nice signal that you scoped things wrongly.
Anthony Kroeger@kr0der
slowly starting to use plan mode a LOT less nowadays i realised whenever i use plan mode, it generates a gigantic plan and then i dont read it and hit build out of laziness having a meaningful conversation with the AI agent to discuss implementation feels a lot easier 🤔
English
finally feeling the AGI. now i just need to allow anonymous cowards to edit shared docs, and i got my own little self-hostable google doc replacement.
it's search will be as disfunctional as OG gdoc search, so not much is lost!
Mario Zechner@badlogicgames
sick and tired, so i can only vibe slop. feeling the agi in my non-crdt, non-ot collaborative markdown editor thing. both opus and gpt should have implemented this: mattweidner.com/2025/05/21/tex… both failed, hard :D
English
Fond memory of my mate’s Chauffeur-Dad helping us do missions using his knowledge of London’s streets
PlayStation Nostalgia@PlayStalgiaX
The Getaway (2002)
English
I’ve had some pretty amazing results with impeccable.style skills
“This app does x
Generate user personas
What does color theory say about this domain and users?
Impeccable:
Overdrive
Polish
Normalize
Typeset”
But granted I don’t think I’ve generated the volume of sites you guys have to have noticed repetitive designs.
I found that back to Programmatic seeded palettes the skills can call, and pre-defined font pairings probably help a lot.
tympanus.net/codrops/2021/1… generating a palette from a random splash image for example to pull the colours away from the defaults.
Also use a much higher temperature in the LLM config when running design creation. Then do a low temperature pass with impeccable normalization skills.
Lots of tokens but crazy results.
English
@sanjaynediyara Yes the models converge to cheesiness, adding a random array of styles is working well now for me
English
All the designs were somehow converging to some brown tones and serif font because I asked for it to NOT do cliche designs
So by that it started doing another cliche, like no sans serif (Arial) fonts, and no typical colors, but then you ended up with brown everything
So now I made a random aesthetics picker like dark cyberpunk, ocean blue, industrial etc. and the designs look more varied like this one:
@levelsio@levelsio
Ok I managed to make the code generated interactive so every idea is really becoming an app now Obvious next question, why not just generate every idea automatically and add Stripe to it and launch them?
English
@thdxr “Guys I built a thing with AI that does (the most basic thing possible), we should be using AI more!”
“Can we spend 500 bucks a month on more automated testing?”
“…no”
English
Jon Sherrard retweetledi
if you are high up at a company you probably miss the days when you coded and shipped features
you feel a crazy amount of exuberance now that coding agents return some of that back to you
it's a huge mistake to think this is somehow relevant to people working day to day
English
@0xSero What about running in sandboxes. Lots of supply chain attacks these days and will probably get worse. Any recommendations?
English
I found the winner the level of polish makes me happy.
Support Droid, Hermes, Codex, Pi, and much much more. Clean design, uses terminals as it should.
github.com/generalaction/…
0xSero@0xSero
Please list all the apps you know of this category that's not already on the list: 1. Cursor Glass 2. Factory Desktop 3. Codex App 4. OpenCode App 5. Claude Desktop 6. CMUX I am looking for this mythic ADE: 1. Has browser 2. Has filesystem 3. Has BYOK 4. Has good ui
English
Offline node module install with an NPM like experience for working without wifi on trains and planes and with poor internet:
github.com/devular/offlin…
English
The new App Store. Coming to your 768GB RAM iPhone 20 in ~2029
@levelsio@levelsio
✨ To inspire more people to go build something now that we have AI to help us (especially non-tech people, cause I still know so many who are scared of building something): I added a [ BUILD IT ] button to IdeasAI.com It's like a mini-Lovable/Replit/v0: Any idea you see you like, you can click [ BUILD IT ], and it will use Opus 4.6 to build a landing page for it And then you can download the code it generated It's not a full startup of course, but a nice preview of what it can be, to give you an idea and inspire you to build it out further The code is live streamed also so you can see it being built 😊 Ironically this itself took me 1 hour to build with AI too Completely free and I pay for the tokens (please don't abuse it :D)
English
@levelsio that was the most fun online / gaming I've had for a very long time
English
@levelsio cool feature but it should show who the link(s) is / are
English