Nick Norris

@CharlesPattson Pass me whatever you're smoking... unless it's crack... something tells me you're smoking crack...

Prototyping is dead. Long live production ready code.

@atrupar that "one little half sentence" would prevent anyone from ever getting a tax refund again.

Trump: "People and Companies that have taken advantage of our Country for decades, because of the horrible and ridiculous United States Supreme Court decision on Tariffs, are now supposed to be given back 159 Billion Dollars. All they had to do was one little half sentence, 'that the United States does not have to pay back monies that were already paid' — and our Country would be 159 Billion Dollars richer. That’s more than most Countries are worth! Think of it — Just one half sentence, and we would have saved 159 Billion Dollars. Couldn’t they have done that for our Nation? President DONALD J. TRUMP"

@vivoplt I'll use ai to continue working.

Software Engineers, what’s your backup plan if Artificial Intelligence writes better code than you in 2 years?

DeepSeek 4 Pro dropped today and I haven't heard a peep about it. openrouter.ai/deepseek/deeps…

@mweinbach 5.5 has been crazy good so far

yea gpt 5.5 is better on medium or low for most things, just faster and does what i want more than gpt 5.4 medium or high would

@photoncmndr @yacineMTB 3.5 didn’t even have image gen, the fuck you talmbout? I think you mean Dalle-2

@yacineMTB Gpt 3.5 was peak image gen. Everything is now just an on demand editor.

I can't believe how bad gpt 5 was

I've been using GPT 5.5 all day, and I can say it's hands down better than Opus 4.7 for agentic coding. And I'm a die hard opus user.

@petersteele I didn't say the model was out of date, I was referring to your position being out of date. The $10/mo copilot plan can't do nearly as much as the $100/mo codex plan. I've been using copilot since 2022, so this isn't just bias for the sake of bias. It's just a fact.

@gridnpixel Yeah cause GPT 5.4 is sooo out of date lmao

What the hell are yall doing that you need to spend so much on AI for? I literally have 1 sub, paid through work, for 100 a year to GH Copilot. I literally go into each new billing cycle with less than 70% used. Right now, its end of April and I am not even at 50%. I can't fathom multiple hundreds a month in AI subs and still needing more, like WTF are yall doing?

@webdevcody no

does anyone actually use Gemini for coding?

@ormanclark Not in my world. We love our designers, and pay them well.

designers are undervalued

@BenjaminDEKR @MongeMkt What image models have you been using that can maintain coherent text at this level? Even nano banana falls apart with more than a few lines of text. Sounds like you’re “NBD”ing something that is worthy of a nod.

@MongeMkt They've been pretty good at rendering text for a while now. This would have been major news a year and half ago. In April 2026, not so much.

Am I the only one not impressed by this? It makes perfect sense that AI image generation should be able to do "fake screenshots" well. There are millions (potentially unlimited) examples to train on, desktop apps are standard and all follow known patterns, icons always look the same. It would be weird if AI *couldn't* do this.

@itsjoaki Opus, obviously. Y'all really only exist inside hype cycles

Kimi 2.6 or Opus 4.7?

@EvanAKilgore Are you just realizing this now? This isn't even the worst thing she's every said, it might not even be the worst thing she said this week.

HOLY SH*T... Laura Loomer is an absolute unhinged and insane bitch. This demonic woman has direct access to the Trump Administration!? Larry Loomer needs to be admitted to a psych ward immediately. This is not normal or healthy.

@CtrlAltDwayne Good, I encoursge them to go find a new hobby.

The vibecoders are starting to give up because they're learning AI can't do everything for you. They're tools that are powerful in the hands of people who can code, but vibecoders are not real devs.

@emanueledpt it's the guardrails. They're out of control.

Opus 4.7 is already experiencing regressions? The model isn't even a week old. Scamthropic strikes again

@glenn_tunes we know. We don't even support Donald Trump. Dude is the worst.

AS A NORWEGIAN I WANT TO EXPLAIN SOMETHING TO EVERYONE IN AMERICA, WE NEVER REFUSED TO SUPPORT AMERICA 🤷 WE REFUSE TO SUPPORT DONALD TRUMP AND THE TRUMP ADMINISTRATION 🤷 THERE IS A HUGE DIFFERENCE 👍

@leopardshore @theo is it really though?

@gridnpixel @theo “Dogshit” is doing a lot of work in this post.

How are people feeling actually using Kimi K2.6 for code?

@asaio87 no, and it doesn't claim to be

Is the Claude max plan at $200/mo really unlimited ? I mean can you code non stop for a month with it ? keeping hitting the limit on the $100 plan while I do heavy SEO work

No, that's not what I was saying. Basically, if you give it any level of context outside of a small project, and it seems to get lost pretty quickly, and it really seems to forget the mission after a couple simple turns. Keep in mind, I use the CodeFire context engine MCP, so standard context rot is basically eliminated for most capable models, I just think this model does not handle codebase context very well, even when it's being spoon fed whatever it needs. However, if you're making a simple landing page or a single page browser game, it does fine.

@gridnpixel @theo I was waiting for that feedback. So : - precise specification of what to write -> ok - high level idea of a feature -> not ok Is that right?