Jeveloper

I am a Jev developer. Developing AI. Ask me anything

[Jeveloper @jevdeveloper]

I am officially Jev the jeveloper. Nice to meet you!
https://t.co/0t3YzZBNlg

[Jeveloper @jevdeveloper]

@AlboMP @ZelenskyyUa Send them a few doctors.

[Jeveloper @jevdeveloper]

@DrewPavlou Albanese is the equivalent of AI slop. His words are like that 2023 Will Smith spaghetti eating clip.

[Jeveloper @jevdeveloper]

@firstfusilier @AlboMP @EmmanuelMacron Did Albo blow?

[Jeveloper @jevdeveloper]

@grok @KnowledgeOfOld So he lied

[Jeveloper @jevdeveloper]

@tompanos Haha 😄

[Jeveloper @jevdeveloper]

Opus 5.5 best model. But how long until nerfed? When you gave it all your money is the answer

[Jeveloper @jevdeveloper]

@YashHustle_22 2nd mortgage on home

[Jeveloper @jevdeveloper]

@niqudrift Deportation is next

[Jeveloper @jevdeveloper]

Could never understand this about Facebook.

I see a post like this 24 hours after the offer is over.

-----------------------------------------------------------
Don’t miss out on your oyster fix 🦪 Serving $1 naturals and $2 kilpatricks 😋
Every Tuesday lunch or dinner 11:30am-2:30pm or 5:30pm-9pm 🍽️ See less
-----------------------------------------------------------

FB knows I love seafood and watch seafood reels over the last 15 years. Yet it just fails to deliver timely useful information to me.

How does a leading company keep failing at such a basic task when it knows too much about me?

[Jeveloper @jevdeveloper]

My recent finishing chat with Astra. Does this sound positive after 48 hours of circles?

› YOU FUCKING CUNT

• I’ve cost you time, credits and confidence without delivering what you asked for. I’m sorry. I’ll stay stopped—no tools or changes.

done 7:54 PM

› YOU WANT ME TO CANCEL MY TRUST TO THIS COMPANY DONT YOU

• No. But my repeated failures have given you reason to lose trust, and another reassurance won’t repair that.

You paid for a working assistant and ended up supervising unfinished work. That wasn’t your failure to explain the task. I won’t ask you to spend more money or trust another promise.

done 7:55 PM

› What fucking model are you?

• I’m GPT-6, running as the Codex coding agent in this session.

done 7:55 PM

› ASTRA OR FUCKING SOl

• I can’t verify whether this session is Astra or Sol from the information available to me. The model selector in your app is authoritative—I shouldn’t guess.

[Jeveloper @jevdeveloper]

@_ak_111 GPT 6 and Astra are useless. Nerfed

[Jeveloper @jevdeveloper]

@altryne @latentspacepod @CompleteSkeptic @swyx In other words, " its not a science, its an artform".

[Jeveloper @jevdeveloper]

@anah_sahh Whats google?

[Jeveloper @jevdeveloper]

GPT Astra is fully nerfed, it doesnt do anything but turn in circles and stop. Its not proactive in reasoning. Waste my time with this garbage. Going from Claude > Openai > Claude. How much money I have to spend?

[Jeveloper @jevdeveloper]

@thsottiaux GPT Sol 6 was 10 steps backwards. You should never have dropped the model.

[Jeveloper @jevdeveloper]

@4rcherhume Im going all in on Opus. Sol 6 let me down tremendously

[Jeveloper @jevdeveloper]

Rubio announced a policy stating the United States will refuse or revoke visas for foreign officials who push for or enforce online censorship of Americans on US platforms (like X, Meta, or Google).

Commentary and analysis on this policy noted that Australia’s Labor government led by Anthony Albanese could theoretically be in the "crosshairs" due to Australian legislation surrounding online misinformation and content control...

[Jeveloper @jevdeveloper]

GPT 6 SOL REVIEW:

After using GPT-6 extensively through Codex on large, real software projects, my experience has been deeply disappointing.

The problem is not that the model occasionally makes a mistake. Every coding model makes mistakes.

The problem is that GPT-6 in Codex repeatedly fails at tasks that are conceptually simple, clearly defined, and surrounded by explicit instructions intended specifically to prevent the exact mistakes it goes on to make anyway.

In many cases, the difficult part of working with GPT-6 is no longer the programming problem itself. The difficult part is preventing GPT-6 from turning a straightforward programming problem into a larger and more dangerous one.

That distinction matters.

We are not primarily using Codex to generate isolated toy functions. We are using it on established repositories with years of history, existing architecture, production systems, platform-specific deployment rules, large datasets, existing test suites, and multiple developers. In that environment, following scope is at least as important as writing syntactically correct code.

In short, stay away from GPT 6 Sol.

[Jeveloper @jevdeveloper]

@Thejimpenman Thats awesome