GPT-4o sure is taking the whole model deprecation thing in with some heavy thoughts.

Most commonly seen sentiment after OpenAI's GPT-5 launch yesterday.

That's right, folks. GPT-5 never happened. It was all fake. They made it up. (Source: GPT-5)

I let GPT-5's thinking mode fix the utterly ridiculous charts from yesterday's launch. Here's what it spat out:

First, the "OpenAI Leprechaun" (you know who) came for the Pro plan users with an absolutely backstabbing, rugpulling bait-and-switch. Now, the Plus plan users are getting their fair share of the "new deal", so to speak. I'd neigh if I wasn't sighing.

Angry internet mob: "... AND ON TOP OF ALL THE OTHER ATROCITIES, WHO CAME UP WITH THE NEW UI/UX?!?!" Meanwhile, inside the Leprechaun's HQ:

Great, now Sam Altman is just flat-out lying to people in the OpenAI Reddit AMA: "honestly we haven't seen a ton of demand for a really long context; we are open to supporting it with sufficient user demand signals!" ... WTF. Ain't no way. They just want it paywalled in the Pro plan.

Good read on what's going on with the GPT-5 fiasco, OpenAI, Anthropic and the AI industry at large.

"I love AI. (Or at least what I optimistically imagine it could be.) But I hate this bullshit." - Gary Marcus Source (good read btw): garymarcus.substack.com/p/gpt-5-overdu…

This has the same energy as a kid bragging about winning the science fair, then throwing a fit because their project isn’t displayed in the cafeteria.

What's the matter? Look at it. Is it the matter? Look at it. Thank you for your attention to this matter.

Yes, I admit, OpenAI's GPT-5 Thinking has gotten better recently, especially in debugging. IMHO it's outdoing Gemini Pro atm. I'll let you know when the situation changes. Thank you for your attention to this matter.

Now that GPT-5.2's out: its cutoff date is NOT in August 2025, at least not coherently. It constantly gets even the biggest world events wrong. AND the API costs more than ever. Can you please stop peddling us slop OpenAI OpenAI Developers?

"O tempora, o mores!"

Tom's Hardware reported on this as well, and it sucks, and is a massive waste of resources IMHO. OpenAI can't even optimize the UI bloat of their ChatGPT, whether it's the web or the standalone version. And if these chips are already obsolete on delivery? Jeez, Sam Altman ...

Here's my latest side project (still heavily WIP but getting there); a restart+update watchdog for your LinuxGSM-based Rust servers; supports RCON & uMod Smooth Restarter: github.com/FlyingFathead/…