@HalkLiff@ForwardEditor But GPT-5.6 Luna Max takes forever to finish compared to GPT-5.6 Sol Medium? Like 3x more time despite having a higher tokens per second? (because it uses much more tokens)
Have you compared the two in this aspect?
@thsottiaux - Why do side threads expire? And you should also be able to come back to them.
- Why is the model unable to "connect" to the in-app browser? Why does this fail often and require a ChatGPT/Codex Desktop app restart?
- Why in-app browser tabs can be tied to only one task/thread?
@thsottiaux I did not appreciate waking up to find that Codex stopped working on the task I ran overnight because of the outrage. But I noticed and appreciated the weekly usage limit top up a few hour ago.
@jxnlco Finally got access after updating the ChatGPT app a few times today, used it for like 10 minutes (mostly testing, figuring how it works) and hit a usage limit.
I have a Pro 20x Subscription.
@athyuttamre Loving GPT-Live in ChatGPT, huge improvement over Advanced Voice Mode.
Small feedback: every time the model says "Checking.." when it supposedly searches the web, PLEASE make it list the sources in the response with a 'sources' button, most of the time it doesn't have it.
@doodlestein Do you tend to use Codex in a vanilla way or do you have many plugins for a comprehensive comprehensive global/system-wide AGENTS.md or a lot of skills, MCP connectors?
Just write the bad sentence.
Pixar's Andrew Stanton on why it's more important to do the work, even if it sucks, than to talk about the great work you'll make (and never do it):
"There's a term for it: whistling on the steps of Carnegie Hall. Most people spend more time telling you the symphony they're going to write than the symphony is written.
Finish the sentence. I don't care if it doesn't have an ending, or a middle. Just freaking write the sentence. It's gonna be bad. You don't get to the nice sentence until you've written the bad sentence. Writing is rewriting."
"It's chipping away at the marble. Every time you decide not to, you're not practicing. So you're just going to be that much less practiced."
@Madisonkanna Iโm sorry, but as an AI language model, I do not have the ability to recognize sarcasm or humor, but it sounds to me like you're describing ChatGPT!
@levelsio Nominal GDP/capita != Quality of life, real purchasing power or affordability/cost-of-living. Means hide inequality (use medians). Regardless diminishing returns kick in hard past ~$60k/GDP/capita. GDP is a flawed proxy. Not arguing for Socialism, just sensible tradeoffs.
@GavinRayDev Beautifully written!
I read the AI disclaimer at the end and I thought "well that is a redundant disclaimer because it's very apparent that a human wrote this, and a talented writer one!".
@doodlestein You are the biggest individual spender that I know of. and 600 commits across 20 projects in a single day is insane. do you know of any other private individual that spends as much as you?
Funny & interesting: did not need to type out all the words of the question to get the answer by Google's AI Overview. Despite the "caveman" wording, it gets it.
"is callstack react poland":
A: Yes,...
@sama@thsottiaux this app UI design NEEDS to be fixed.
Sorry for the language in the image it was because this 1000th time I got really angry for some reason.
Attached are images/screenshots of what I'm talking about and my recommend UI design change.
@diegoasua@TheHumanoidHub@alexocheema Looks like the new Jetson delivers more TOPS per dollar than the M4 Mac mini. What do you think? is it worth it for you to experiment with these single-board computers (that are similarly sized)?