WholeEntertainment

[WholeEntertainment @WholeEntertainm]

@jun_song The correct view is that AI is still too expensive at the moment compared to the value produced.

[WholeEntertainment @WholeEntertainm]

@gonifacundo @adamlyttleapps No nerfs. There were problems and they were solved.

[WholeEntertainment @WholeEntertainm]

@Dex1126155 @wholyv The model and technology are so new that we are literally the beta testers.

[WholeEntertainment @WholeEntertainm]

@melvindvivas Luna Max is slow so this %/h comparison is misleading.

[WholeEntertainment @WholeEntertainm]

@thsottiaux Did you mess with follow-ups?

[WholeEntertainment @WholeEntertainm]

@cnyngyuyng1 @thsottiaux @nudevise I'm not anyone's representative. You have your legs, I hope

[WholeEntertainment @WholeEntertainm]

@MatthewBerman Parallel Trading Bots

[WholeEntertainment @WholeEntertainm]

@OpenAIDevs It would take a reset to apply it to all projects

[WholeEntertainment @WholeEntertainm]

@MatthewBerman @bot @SpaceXAI Parallel Trading Bots

[WholeEntertainment @WholeEntertainm]

@akshdeeps_001 LOGOS≠OUTPUT

[WholeEntertainment @WholeEntertainm]

@Im_IrushiK If GPT-6SOL consumes so many limits, OpenAI usage has a problem. Aren't they that silly?

[WholeEntertainment @WholeEntertainm]

@buildwithrajath All Eyes on Grok

[WholeEntertainment @WholeEntertainm]

In Rome we’d say “grazie al cazzo” — roughly, “no fucking shit,” except with the added implication that what you just said is so painfully obvious even my dick could’ve put it better.
You’re comparing a Ferrari performing like a Ferrari to a kick scooter somehow performing like a Ferrari.

[WholeEntertainment @WholeEntertainm]

@shredbuilds No worries!

[WholeEntertainment @WholeEntertainm]

That’s all true, but the level of effort here is clearly different. Astra probably interpreted the prompt as if the animation on the right were a simpler problem than the one on the left, and therefore allocated less reasoning to it.

They may have lowered the juice. With a better or more structured prompt, you could probably get the same result. But I think that’s exactly the interesting part: given the same prompt, Astra interpreted it differently.

[WholeEntertainment @WholeEntertainm]

@AmortisX @rezoundous No, it's not. Either you don't know what I'm talking about or you're consciously lying about. It's also unusable on the x20 plan, I used 4 reset in less than 24 hours and I'm at 0% for the next 6 days

[Rajath Gowda @buildwithrajath]

🚨 GPT-Astra is making everyone lose their fucking minds. 👀

- Unpredictable limits.
- Unpredictable usage.
- Unpredictable reasoning.

One day Astra feels like magic.

The next day you're staring at your usage meter wondering what the hell happened. 😭

The model is insane.

The product experience is chaos. 👀

[Lxyv @wholyv]

the recent nerf observed in gpt astra could possibly mean one of the following:

OpenAI is about to launch GPT-6 Sol since the model already appeared in arena. they will most likely time it with the next Claude model, which is expected to release soon.

Compute shortage might’ve led them to mess with juice values of the model. I have personally observed Astra using less usage as well than it normally did.

[WholeEntertainment @WholeEntertainm]

@totoyayoo @buildwithhassan I would place it in the section "Propaganda versus open-source"

[WholeEntertainment @WholeEntertainm]

@totoyayoo @buildwithhassan I agree with you that it's unrealistic, it's probably just hype. This does not mean that it is not feasible with the right filters/harness