Builder. Looking for unique creations to spotlight.
Based in Estonia
- Tweets 516
- Following 456
- Followers 164
- Likes 1,292
GPT-6.1 Astra is not coming out after all.
Sol 6.1 Ultrafast was the product being hyped by the OpenAI team.
I don’t know why anyone would be excited for this release.
Sure Sol is slow but who would want to pay $500 a month to make it faster.
It looks like Fable 5.5, Opus 5.5 and Sonnet 5.5 will not have competition anytime soon.
41
8
277
14,605
Maybe normie users are the way to go? I think power users will bring anthropic the good data to make their models stronger
1
1
12
GPT paced the frontier by giving out a model that now runs at the pace of a M4 Mac mini running local llm while telling you that you'll be 5x more productive
It's a masterclass
I can't comment on Dots because it's 2026 and the supposed people who are going to herald in a new era for an agentic an connected humanity can't figure out how to fairly distribute AI to beyond America
1
3
85
Gemini 4 Argon is insane
I know literally NOTHING about coding. ZERO. and i just built 3 fully functioning web apps in 30 minutes
http://localhost:3000/
http://localhost:5000/
http://localhost:8000/
check it out
25
4
94
9,426
chatGPT 6.1 solのウルトラでこのクオリティ？？笑
もっとリアルに作れるのかと思ってたから絶望笑
いい感じのこと考えたんだけど作れなそう、、
人を作りたいんですけどどうやったら作れますか！！
教えて欲しいです
39
10
179
50,573
😂 kinda funny though..
is that blender?
try @HumGen3D which gets you started with a proper body rig
1
74
The bot replies to every single tweet are out of control. Just LLM slop everywhere. And like… to what end? Why?
53
8
649
31,217
My feed is full of codex users all noticing that after today’s reset they burn through their limits absurdly fast.
I’m on the $500 plan and burnt through 10% with two Astra high agents in 30 mins. Whats going on @thsottiaux?
18
1
32
3,769
Over the last 24 hours, GPT 6.1 has been roughly 20 tok/s
But some periods have been as slow as 11.6 tok/s
The fastest was 28 tok/s
What are you getting?
1
1
68
Instead of dropping vague posts and cryptic comments in random contexts, OpenAI should just pick one of the posts that clearly lays out their biggest problems, mine or anyone else’s, and say:
“We see this. We’re working on it.”
That would make people a lot happier than all the little mystery games.
No secrets. No vague hype. Just acknowledge the problems and fix them. 😐
21
4
113
3,819
Nvidia have had massive pressure on their hardware since crypto mining.
Like Apple, they built their position by making products enthusiasts loved. Then they realised they could charge 2–3x+ and people would still pay it.
There’s very little incentive for them to commoditize their hardware now.
We need new providers willing to pretend they’re building for enthusiasts long enough to become the next Nvidia or Apple.
Anyone else disgusted with the Spark announcements today? Might be putting the two I have for sale soon. Looking at alternatives now since any efforts by enthusiasts is being cut off at the knees.
I admit it… I was duped by them. Sold off my workshop equipment for? Might sell and use the $$ for China based inference. People can say what they want but at least those companies are enabling individual enthusiasts to build .
Definitely will not be buying an RTX Spark laptop in November and supporting this moving forward
1
2
114
Gpt 6.1 it's unlimited
You'll never run out of usage
Only caveat is that it takes 3 hours to get a single turn.
3
4
105
It’s genius if you think about it. Usage feels huge and slow responses keep servers from melting
1
2
36
OpenAI is going to release GPT-6 Bel.
Fable 5.5 drop will happen any day now and it will mog every model.
Anthropic will then have Fable 5.5, Opus 5.5 and Sonnet 5.5 at the top of every leaderboard.
Haiku 5.5 is a wildcard and even it could be an absolute monster.
Codex will have to respond with Bel and have an Astra moment.
The questions is does Claude have a model that can mog Bel?
152
60
2,003
250,625
Bro, it's the majority. We on X live in an ecochamber while people out there ask ChatGPT(yes, the UI) for code and paste it
1
3
160
So true.. but I have fond memories of copying and pasting from chatgpt.com
1
3
105