Your Product has never been judged. It has only been tolerated.
https://t.co/w9qcLlo30c changes that.
AI inspects your site and publishes a verdict you can’t edit or appeal.
It judged itself first:
“clear but visually thin.”
Check, be judge , have fun. Lets goooo.
Roughly 7.05 hours of continuous GPT-6 Astra usage at X-High reasoning, on a single thread, with the $100 plan.
Choose the plan that works best for you.
Good job, @OpenAI 👏. We are Doomed.
Welcome to AGI world.
I was on the $20 plan during GPT-5.5.
Surprisingly, GPT-5.5 High was excellent from a usage-limit perspective. In a 5-hour window, I could use it heavily for 4–4.5 hours, and my weekly quota would still last around 3–4 days.
Roughly 40–45 hours of serious usage per week.
Because I kept hitting the weekly limit, I upgraded to the $100 Pro plan about 4–5 days before GPT-5.6 launched.
And during those few days?I used it like crazy. Sometimes almost 24 hours continuously.
Still, I was only consuming roughly 10–15% of my quota per day.
Then GPT-5.6 arrived.
Everything changed. Suddenly the same $100 plan wasn't enough anymore. My quota started disappearing in 2-3 days of what I would consider normal usage. It felt 2–3× faster than GPT-5.5.
This was especially confusing because GPT-5.6 was supposed to be a more efficient model as both have same api price.
Then GPT-6 arrived. Now the quota can disappear in roughly 10–15 hours of serious work.
Yes, GPT-6 costs significantly more to run. But if newer generations are also becoming substantially more efficient, where are users actually seeing those efficiency gains?
So I upgraded again.
$20 → $100 → $200 Pro Max.
And somehow I'm now in a situation where even the $200 plan can run out in around 1–1.5 days of work when using a single Astra Max workflow. And have to wait. Sometimes several days. Sometimes until the weekly reset.
So congratulations @OpenAI 😂
You successfully converted me from a $20/month customer into a $200/month customer...
...and somehow I spend MORE time worrying about quota than I did on the $20 plan.
And maybe all those generous resets we used to get weren't really "free" after all. Everything has a cost. Perhaps we're seeing that cost now.
I just have one question:
Where are all the efficiency gains going?
Astra generated this Taj Mahal 3D model using Blender MCP—and then created and rendered the entire walkthrough video inside Blender too.
Agentic 3D creation is getting seriously powerful. 🤯
Restarting development on a project that brings Remotion and HyperFrames under one engine.
The UI is still shit.
But I’m working on it 😅 with Astra
Want to help build it? Contributions and feedback are welcome. Check it out, and drop a ⭐ if you like the idea.
https://t.co/cWp8z2rrlN
Has anyone watched the trailer for Artificial, the movie about Sam Altman and @OpenAI’s journey?
Curious what everyone thinks 👀
https://t.co/I8d6YcdkBm