@Adam_lag_ Yes obviously you will pay for both that's the entire point, to see how they're doing so you can select the best one for your task for the future
@Adam_lag_@testingcatalog What do you mean? The point is to compare 2 models for the same prompt and see how they each do. Very useful. Like compare Opus 5 xhigh to Fable 5 medium, no memory.
Claude Code on desktop now has an in-app browser.
Claude can pull up docs, designs, or any other site. It can read, click through, and interact the same way it does with your local dev servers.
It's sandboxed and configurable: you choose whether sessions persist.
@80sGeek@AndrewCurran_ This does not make the model dumber, not sure how you've learned this, but it's not true. They were already using watermarked text, they just need to change it. Nothing major, won't affect the users.
@DanDr1s Because giving banked resets means people actually use the entire usage twice in a week. While with ther random resets, just a few people benefit. Yesterday the normal weekly reset happened and a few hours after that, Tibo decided to reset, i got no usage at all from it
@mirochill That's how they've been doing it lately, users don't actualy benefit of it, it's just for PR, not for real users benefit, almost like a fraud. They promised we'll get banked resets so we can reset whenever we want, but they keep resetting when users don't benefit from it
@DanDr1s Ofcourse, except their "resets" were misleading/a a fraud (just moving the weekly reset period) plus a lot of them came just hours before the planned weekly reset. They rarely gave actual, real "banked resets" that you could use when your real usage was depleted
@shariqriazzz@testingcatalog Yeah i know very well, see my other comments in the thread. The socalled "resets" were mostly a fraud due to this. Hopefully these paid resets won't be more than 10% of a sub, it they ever release them - I would like this to happen personally
@IsaacHinman@testingcatalog I saw it too and i have it disabled in Codex. Even had Codex do a verification in the past sessions and none had auto approve enabled. This is messed up
@shariqriazzz@testingcatalog Prices are too high indeed but these are also "banked resets" so true double usage in a week. A 25% price would be fair for one extra week worth of usage (better than the alternative of buying a full second subscription)
@testingcatalog I doubt that more than 10% of the Codex users deplete their usage. Tibo himself said that a small percentage of the users are engaging in long horizon coding tasks the kind that maxes usage. The rest (90% presumably) of the users don't even benefit from the so called "resets"
@testingcatalog Prices are too high and your estimates are wrong too. Remember that they do not reset the usage inside a week, they start a new week instead and what was previously the weekly reset then no longer happens
@scaling01 This is just wishful thinking. He had a lot of "projects" like this, just PR stunts, never materialised. Note he won't use his trillionaire money for anything, it's basically he telling the investors to invest their money.
@AndrewCurran_ Wow, so original! I bet they even brainstormed, "how do should we name our company, must be something original, unique" -> and then someone said "energy"...
@Ixel111@btibor91 It's a weekly reset. Remains to be seen if they want to target 25% of our subscription price, or less say 10% (considering we got like 13 free resets in july)
@bebraSOL@DanDr1s Anthropic is the one holding the biggest revenue, then OpenAI, but we don't know much about Google since they don't have pure AI plans but general combined Google services subscriptions
@DanDr1s For that, they need to have the mainstream people actually pay for AI. Currently it's just a tiny fraction of the total billion+ users that is paying, like 10-20 mil subscribers, we're in the 1% paying now, they'd need to increase that tenfold while also making companies pay more
@scaling01 10M users and it feels like a low quality vibecoded app. Very long loading times when loading a conversation even on a highe end pc and stutters like crazy. I think what happens is they refuse to add loading indicators while the sandbox gets created/recreated