thatscypher

@goonernumberuno both my parents had a stroke because of that

@thatscypher im super healthy i’ve been smoking since i was 13, ill be fine 💕

marlboro red is better then marlboro gold who agrees with me

@sofakight steal from a bank 👍

where to get 18,000 dollars before I take out student loans🫩

@Layla_3D what happened to the topology 😭

How it started vs How it ended I modeled every armor element with its own physical height and depth!

@sama I’d agree if I didn’t have to sell a kidney to use it haha

5.5 xhigh in fast mode is really good i think i got psyoped by twitter on medium for a bit

@haridigresses Low is very fucking unreliable for being the best

GPT-5.5 Low (specifically Low) is the best coding model I’ve ever used, and it’s not particularly close.

@HarshithLucky3 the model doesn’t know these things, if you know anything about how LLM systems work… its just hallucinating answers based off some value, but the thinking is measured by tokens not fucking numbers

Juice values of GPT 5.5 in Codex low - 12 medium - 24 high - 96 extra high - 768 same for standard and speed

@thsottiaux @nima_ab What’s usage limit? I’m a Codex user not familiar with these terms

@nima_ab Did you know that you've hit your usage limit?

@thsottiaux CoD: Ex

Please don't pronounce Codex, COD-ex. We are also not a fish.

@GeorgeCrudo yep, it’s just an overpriced python wrapper, you pay 15$ a day to let a machine do something thats doable with 2 lines of code

A handful of replies to this are "you're using it wrong. it's actually good for doing batch commands across tons of scene objects" and I'm begging you guys to learn some basic scripting because that is literally all that is. The AI / MCP is not giving you this capability.

@sudoingX in less with pro plans you can make something that buys you 100x 5090s

fun fact: 4 years of $20/mo to chatgpt = the price of a used 3090. one leaves you with nothing. the other leaves you with a 3090.

@Elaina43114880 you’re on 2 fucking percent battery and this is the best question you can ask it…

The juice value of GPT-5.5 Thinking Extended in ChatGPT is 192. The prompt can be copied directly from the comment. What is yours? Please share it in the comments 👀

@rainbowslop u lost ur jeans?

@rainbowslop u missed

@sama what sam? also pls reset limits ty need codexmaxxing but my 2x 200$ plans aren't enough

big upgrade for codex today! try it for non-coding computer work.

@emanueledpt letting it work through 7 planned PRs tonight, should be a 5-6 hr work, has reviewers setup and knows how to work on my PRs. It’s a massive refactor / alignment in my project curious to see the result (or if it hallucinated and deleted system32)

I’ve tried 5.5 for a couple of days now And this is such a good model, if you grasp how to use it It’s different from 5.4 What I noticed is that it is more prone to action than previous models It has more personality too I always have it set up in Low (never thought about using a model in Low, but this is so OP) If I need more horsepower in High, haven’t touched xhigh yet But this model just works Team really cooked @romainhuet @ajambrosino @dkundel @thsottiaux

@Angaisb_ this is instant?

I'm liking GPT-5.5 so far but it definitely still needs personality improvements

@iruletheworldmo if it doesn’t happen?

this will go down as one of the biggest days in human history. people will finally start to wake up to the massive transformation that this technology will bring. ai is here to stay, and the abilities of this model confirm it we are on an incredibly steep trajectory and this is the first real taste of just how quickly this is all moving i want you to remember after you try it today this is not the end, it will sharply increase in intelligence and usefulness every 1/2 months from here until the end of the year beyond that, it's impossible to see so let's walk through the mist together and ensure we capture the beauty of these systems and minimize the downsides

@0xSero anthropic is just that bad

It really is just Claude Code sucking. Been cranking Opus-4.7 via Droid's API since it's discounted and feels like a different model. It found a way to write an observer that patches Codex app at launch to inject all my models from Vibeproxy. I can even use Claude in Codex

@kr0der Tibo got you

guys i'm not gonna make it, 34% with 6 days remaining