ThoughtMaxer

Working with every frontier AI model 10-16 hours a day at scale. Let's bring on the singularity!

[ThoughtMaxer @ThoughtMaxer]

Your frontier model lags behind Astra Ultra in coding, and plans have 5-hour limits at any price. I won't compare it to Opus 5.5, as that's several steps ahead of Wonder. So there's no compelling reason to use it. I keep a $50 account to test the new stuff as you release it. It is sitting dormant right now.

[ThoughtMaxer @ThoughtMaxer]

@elonmusk ROFL Grok and several fleets of accountants and financial planners and managers.

[ThoughtMaxer @ThoughtMaxer]

@imagine @WonderStudiosX @glumlot_ Almost there, but it's still clearly AI and inferior to human actors.

[ThoughtMaxer @ThoughtMaxer]

@Officially_Dev_ ChatGPT is the most used app in human history. That's how😆

[ThoughtMaxer @ThoughtMaxer]

@Da7_Tech Ask yourself, what are you most hyped for? Distilled GPT Astra (Kimi/Qwen) you've already got in your ChatGPT sub, or the new frontier?

[ThoughtMaxer @ThoughtMaxer]

@thepaulthomson Yeah bro, need ship! Opus be mogging you bad.

[ThoughtMaxer @ThoughtMaxer]

@codex_resets Any improvements to sessions and reasoning are KEY. No reset.

[ThoughtMaxer @ThoughtMaxer]

@thsottiaux Decent ship. Nobody ask for a reset!

[ThoughtMaxer @ThoughtMaxer]

@TokenGremlin They’ll be “merging” with someone somewhere (aka getting bought out) very soon if they can’t beat opus 5.5. There’s no justification for their existence if their models are consistently beaten on cost and ability.

[ThoughtMaxer @ThoughtMaxer]

@ArtificialAnlys @Kimi_Moonshot Yeah but where do they rank on the coding agents benchmark? That is all that matters.

[ThoughtMaxer @ThoughtMaxer]

@omkarships It was never alive in the first place 😆 - Actual Chatgpt fan

[ThoughtMaxer @ThoughtMaxer]

@matthias_mrc You must not be building anything very complicated. I get a lot of "that's on me" all day from all the frontier models

[ThoughtMaxer @ThoughtMaxer]

No waste of time in this line of work. Just gaining knowledge😉 Honestly, I don't distrust Elon as much as others. The Grok bot is the king of bots right now and it is a needed functionality to do AI at scale. Like I said, I want to see how Opus-Duet handle it and I might up my Opus capacity if it goes well. Meta is a different story. The $50 Muse account sits unused. The latest Muse was one step behind Astra & even Sol 6.1, and they had 5-hour limits. Ran 2 short jobs on it and then it went back on the shelf waiting for Meta's next "frontier model" announcement.

[ThoughtMaxer @ThoughtMaxer]

Did my experiment. It basically just opens a Codex chat to do the work. It's a nice little helper, but I could have done that myself, and I had to intervene in the chat anyway. This is light-years away from the Grok team of 18 bots I had the main Grok create that literally help run ops on this server farm. Honestly, this product is a bit of a disappointment from OAI, and I say that as a ChatGPT main (though I use them all constantly.) Can't wait to see what Duet does once it comes to wide release.

[ThoughtMaxer @ThoughtMaxer]

@joshuakirchne17 @thsottiaux Right now, in the middle of drafting a protocol to get it working. OAI DOES NOT make it easy lol

You've inspired me to try out my spare Dot capacity.

[ThoughtMaxer @ThoughtMaxer]

@hooch133 @KillaKreww Looked to be doing fine there

[ThoughtMaxer @ThoughtMaxer]

@thsottiaux I'm out of usage with this new codex regime. Tibo is day 17 a reset since there was nothing announced?

[ThoughtMaxer @ThoughtMaxer]

@billgardinerx @infodexx 0.4% of all Indians live in the US. 25% of humanity is of Indian ethnicity/from the Indian subcontinent. So they DO live there dude lol. A significant % of Indians do not live in the US or even outside India.

[ThoughtMaxer @ThoughtMaxer]

It is not about security prompts. I've never gotten one either. It is about security theater, meaning it'll create so many checks that it won't be able to proceed because of their weight. And most of them would be meaningless. I had Astra create a chat that was supposed to "prepare to deploy," but when I told it to deploy, it refused because "this chat is supposed to only prepare to deploy." It then gave me the exact language it needed to start deploying. I posted that in, and it goes, "Yes, I have that now, and yet this chat is to prepare to deploy, and thus I can't deploy"!! Now that was just an Astra chat. I don't know if the dots behave that way, but when they first came out, I tried to put mine on agentmail, and it refused to take any instruction unless I typed it in. So that's the kind of stuff I'm talking about. Damn, that's a long post for X. Sorry😀

[ThoughtMaxer @ThoughtMaxer]

@alx Kara Swisher is truly out of her mind. That said, I'm glad Elon is watching this closely.