Pinned Tweet
The only thing that can't be a wrapper is how you want the world to be. Act on it
Forget everything else is most likely a wrapper.
27
We are now in the ai marketing era.
HELP WE NEED FOUNDATIONAL STUFF LIKE WHAT IA THE LINUX OF AI.
I hate to break it to everyone, but Opus 5.5 has been nerfed starting the last few hours today.
Autistic Claude speak is back, it's doing really dumb shit & this is not the same model I was using the last 2 days.
Confirmed this on multiple accounts, it's turned into an idiot.
4
ƒAmalgama() retweeted
I'm really disappointed in OpenAI, recently codex has gotten way less reliable and lots of other issues.
First, I thought they "had compute" and now 20x plans are still paused while Anthropic the company with "not enough compute" lets people still buy them.
Second, I have multiple codex accounts and each single one of them keeps facing problems. I run agents for long periods of times, and it sucks when I come back to a error saying "Servers overloaded" and my entire agent gets paused. This has gotten to the point where at the worst times (around 10pm ET) I'm receiving this every 5-10 minutes.
Third, GPT 6 Sol is disappointing. It's either nerfed or something else is up, as this model can't understand basic things I say. And before the GPT 6 Sol & Luna launch, Astra got nerfed to the ground. Astra basically couldn't do anything meaningful.
Fourth, Anthropic gives better limits. I never thought I'd say this but Anthropic gives you higher limits than OpenAI if you use Opus 5.5. GPT 6 Sol burns through my limits 3x faster than Opus 5.5 does.
@thsottiaux I don't know what is happening at OpenAI but this is not a good look, the entire community likes you and the team at OpenAI. But this is really not ok, we need some clarification, fixes to these issues, and resets to compensate.
48
27
539
17,858
This is the bitter lesson of harnesses, you are a slave to the ai that is in the harness.
If the new models goes to shit your only options are:
1) Endure the pain because everything is with that company.
2) Change all to the company that has the good model for the moment.
3) Get a harness that is not dependant on one ai model and that works well like codex
Crazy to say but point 3 is still for grabs like what is the go-to harness that is not Claude code or codex that is really good?
I don't see a clear answer.
$500/month is the wrong move by OpenAI
That is literally rent money for the third world
More importantly, it signals real compute scarcity or the next model is strong but inefficient.
Astra med consumes in a day what Opus 5.5 ultra does at 10% of weekly usage.
Where compute?
13
Look at this
AFK Jev Test Stream | My JevJev is becoming Ironman, on the Road to Vostok! nitter.net/i/broadcasts/1OGwbnVRg…
15
Funny enough I think they are actually right on this one but who uses blockbuster anymore they found the model for movies but the industry out model them, why wouldn't it happen with rabbit.
Rabbit has actually the opportunity but I think they don't have the moat to hold, they need to become a lab and build like a rabbit Linux omachy wrapper.
Something more than been in the right place at the right time.
15
This is moat is so true but how are these companies?
.@vkhosla's prediction for the two moats of personal AI:
1. A company you can trust with your data
2. A tool that actually gets the job done
“I believe the moat will be trust in a company that develops trust with consumers. And that’s where Meta has a huge disadvantage.”
“I also think completion is going to be really important. Consumers will be loyal to products that get their shit done."
“If you can get all my stuff done, I’ll be loyal to you.”
13
This sounds like they made muse different but it's just open claw with ai model train for task oriented workflow (goal on steroids) plus the data extraction that goes directly to meta
use muse to save yourself all the time of checking different barbershop websites, calling for appointment times, checking google reviews...
muse can do all of that for you!
34
Hahaha omg did you really think they where giving you a OpenClaw wrapper for free.
Hahahahahahahahahahahahahahahahahahhahahahahahahahahah.
‼️ FYI: Meta's Muse opts your data into AI training by default
The app just hit #1 on the App Store. Congrats to everyone who downloaded it and immediately became unpaid AI trainers.
As reported by Wired, Muse continues Meta's proud tradition of deciding your data is theirs until you say otherwise. And this time, it's not just your posts: it's nudging you to hand over your bank account, email, and passport. You know, for the AI.
Here's how to un-volunteer yourself:
→ Go to Settings
→ Find "Data Controls"
→ Toggle off "Help improve our AI models"
(Translation: toggle off "let us profit from your personal information without asking nicely")
Remember, it's your data, not theirs.
12
🧊 is really music to be able to fight any Goliath moment in your life. @Drake
open.spotify.com/album/0OAv7…
52
Raising billions to build AI by convincing everyone that only AI controlled by us can prevent AI doomsday
76
This is like saying you just have to close your eyes not to see it.
Isn't this denial, is like when a boy becomes man it's just different time.
That is not a bad thing.
20
