Miłosz Karolczyk

Product Engineering Tech Lead • Digital Transformation • Software Development Strategy • Software Architecture • Domain-Driven Design • AI

[Miłosz Karolczyk @mkarolczyk_dev]

Hello everyone! Let us know what system architecture is mainly used in your projects.

#architecture #monolith #microservices

[Miłosz Karolczyk @mkarolczyk_dev]

Dokładnie...

1/ 📍 Październik 2025 NYT: widywani razem towarzysko. To pierwsze publiczne powiązanie obu osób. W tle federalne śledztwo wobec Megaspeed w sprawie przemytu chipów Nvidii do Chin.

2/ 📍 Koniec 2025 (niejawnie) FBI zgłasza kongresmenom obawy, że Alice Huang może przekazywać informacje Chinom. Wg NYT o relacji zostaje poinformowany Trump.

3/ 📍 Maj 2026 13.05: fundusz Culper Research (short seller) publikuje 40-stronicowy raport o „niekomfortowo bliskiej" relacji. W tym samym czasie w administracji krąży tajna notatka nadzorowana przez Rubio (wg Axios).

...

[Miłosz Karolczyk @mkarolczyk_dev]

@thsottiaux Improve the current limits; they're at least 3 times worse than the competition's.

[Miłosz Karolczyk @mkarolczyk_dev]

@thsottiaux Poland's Constitutional Tribunal needs a reset. Will you help?

[Miłosz Karolczyk @mkarolczyk_dev]

I managed to get Dots working once while I was outside the EU. I normally live in the EU, so Dots doesn't work for me. Despite that, I get several notifications from ChatGPT on my iPad every day that I can't read in full. 🤦

@thsottiaux @OpenAIDevs

[Rikiya Takehi @rikiyatakehi]

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.

We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: https://t.co/bOWCcFhCVq

[Miłosz Karolczyk @mkarolczyk_dev]

@theo Thank you, this is the option I've been waiting for! It was the only thing missing!

[Miłosz Karolczyk @mkarolczyk_dev]

Terra is really good at everyday tasks and very fast, too. Check out how quickly and effectively it handles browser interactions. I've been waiting for a model like this for a few of my long-running tasks.

Now the only need 2x usage limits across the board ;)

@OpenAIDevs thx!

[Miłosz Karolczyk @mkarolczyk_dev]

@Ionkosm OpenAI now has exceptionally low usage limits on its subscription plans. Those limits are precisely what prompted me to return to Claude.

[Miłosz Karolczyk @mkarolczyk_dev]

@RobertJBye That would be wonderful!

[Miłosz Karolczyk @mkarolczyk_dev]

@thsottiaux , you recently said that you only need to activate Dots once outside the EU, and they'll keep working afterward. Challenge accepted. I actually traveled outside the EU to try it. Now I'm back in the EU, and Dots have disappeared…

[Miłosz Karolczyk @mkarolczyk_dev]

@BrendanPlayford @thsottiaux I feel the same way. I love OpenAI, but I'm fed up with the constant experiments and moving goalposts. Fix this and don't give us reasons to leave.

[Miłosz Karolczyk @mkarolczyk_dev]

@thsottiaux Additionally, I keep getting 429 errors from Luna when trying to invoke subagents.

[Miłosz Karolczyk @mkarolczyk_dev]

Something is seriously wrong with Codex limits over the past 2 days. @thsottiaux what did you break?

I run the same process every day, but for the past 2 days, I've been burning through my quota much faster, and there's no reasonable explanation for it.

[Miłosz Karolczyk @mkarolczyk_dev]

@thsottiaux Less of this. It's impossible to work today; this message keeps popping up.

[Miłosz Karolczyk @mkarolczyk_dev]

@wholyv It's just that most of the people who voted had Plus, not Pro.

[Miłosz Karolczyk @mkarolczyk_dev]

@ChatGPT When will Dots be available in the EU?

[Miłosz Karolczyk @mkarolczyk_dev]

@Dimillian If it can read minds, why doesn't it just implement everything at once?

[Miłosz Karolczyk @mkarolczyk_dev]

@OpenAIDevs , wake up! You're losing touch with your audience!

[Miłosz Karolczyk @mkarolczyk_dev]

Exactly, OpenAI has become a childish, unserious company. We expect stability and predictable limits. That’s why many of us trusted OpenAI, but now that trust is crumbling.

[Rajath Gowda @buildwithrajath]

🚨 I think these “free” Codex resets are creating a bigger problem for users than OpenAI realizes.

I’m seeing A LOT of people pissed about these global resets.

And honestly, I get it.

Imagine you finally decide to use your banked reset.
You barely use it.
90% of your usage is still sitting there.

Then suddenly…
GLOBAL RESET. 😭

There goes the 90% you were saving.
And now your next reset date can get pushed back too.

WTF?

If you're going to reset everyone, at least give us a fucking choice:
→ Give us banked resets
→ Give us 24 hours window to use our existing usage
→ Don't push our next reset date back another week
→ Don't wipe out usage we already have

We don't need surprise resets.

We just need control over the usage we're already paying for.