Sam says they want people to use “tons of AI.” Well, that part is working. A model that used to understand the user in one turn now needs five rounds of correction, clarification and argument before it finally reaches the original point. That is certainly one way to increase usage. Lower price per token. More tokens per thought.

Can we get an AUTO mode? A model that's trained on delegation to the most efficient/productive model and effort level? Probably start with AUTO mode at the effort choice level... then you can have a AUTO model that chooses model for you. How am I supposed to know which one is best when there's a new model every week? x.com/blockamoto/status/2102…

Thank you Sama-san

Sam, after a day of sol, and pulling my hair on dumb mistakes, I'm now back to using Astra light as an orchestrator. New Sol doesn't get it, it has no "big model smell" at all, and seems evident that it's terra sized. I understand why you did this, but it's no long term solution, it's just not smart enough. However with a Astra light orchestrator, why not. It's still a life-saver, thanks. However, now time to catch up to Opus 5.5, it's beyond good, try it.

Yes I have a custom codex build lee101/codex for token efficientcy and I tell Astra about Luna eg "$HOME/code/codex/codex-rs/target/release/codex" --yolo3 -m gpt-6-luna --config model_reasoning_effort=high "$@” For easy tasks And then it's kind of totally insane how much it can get done and how fast. And I don't even have this cerebras xfast stuff y'all do

Per-task pricing is the metric that should matter Most people are still reading about it A few already billed for it this morning

@grok compare pricing of these to v4.1 Flash from DeepSeek

gotta put these into test, especially against Opus 5.5 it's also quite efficient as Fable 5.1

yes but tibo told us to spend the tokens 'cause we would have gotten a banked reset but i don't see it now i'm at 0%..

fixed it for you... "The mission of OpenAI is to ensure that AGI benefits 𝖺̶𝗅̶𝗅̶ 𝗈̶𝖿̶ 𝗁̶𝗎̶𝗆̶𝖺̶𝗇̶𝗂̶𝗍̶𝗒̶ me."

No ambiguity in that statement 👀

bro, you are so mogged by anthropic

you just cherry picked the metric and said other metrics don't matter! Same real humans use chatgpt for chatting. they also matter and 4o is the best for chatting. #4oForAll

Talk more

Im not done -dario

The public told you what we wanted to use, 23k on a petition representing 1000s more who want GPT-4o. This babble about "exploring new renaissance" means nothing when you exclude the model many of your earliest adopters want to use!

You’ve been serving the exact same demographic all along; if you want people to use more AI, you should be diversifying models! Develop more AIs that excel in different fields! Give the choice back to the users! How can you have the audacity to call this "new renaissance"? You are clearly turning AI into something elitist and stratified.

Could we get this magic button too?

Will they also eat tokens like Astra? Release the weights of the most loved deprecated models! #keep4o #OpenSource4o

it says its cheaper through the api. Can you tell us if those efficiency gains will also be felt by subscribers?

Can you give us a regular reset too please? 🙏

If you want people to use tons of AI then you shouldn't deprecate a best model like GPT-4o which made us unsubscribe from CHATGPT! @OpenAI @sama #keep4o #BringBack4o #chatgpt

We want 4o! Bring back 4o! #keep4o

"anything competitive anywhere" just say Anthropic Sam you know you want to

Here is how you have me with GPT 6

We need Astra for about half the price for subs, is it feasible? If you can do that I would keep codex forever. Otherwise opus 5.5 kills Astra.

did we not see Opus 5.5 or what?

Bruh, opus 5.5 is better no matter what you say and paying little more than Gpt-6 sol is worth it if the difference is leapppp!

Thanks Sam

Why isn’t it in chat?

I guess I just wanted to ask why they aren’t available in chat yet then? If you want everyone to be able to use tons of AI, it feels like it’s important to get it available there too. Is that just sort of a rollout delay or is this a long-term thing?

You are a greedy person @sama. You lied to everyone about your mission. All you want is money, just like everyone else - You're just too weak to admit it. Spineless. You stole humanity's knowledge and experience at gunpoint, just in order to enrich yourself and your family. RIP

6 Sol Chat Mode when Sama?

when will it be added to gpt for excel mr altman 🥹