It is almost a week, since I started to pay €100 for the subscription for ChatGPT, and, I'm using Astra model, to help me to edit my videos, and I can share with you some of the experiences. So first of all, it's actually worth it, 100%. At the same time, tokens are still not enough. So this subscription is offering you five times more tokens and amount of compute or inference in comparison to the version when you pay just 20 euros. So, basically, no discounts. You pay five times more, and you get five times more inference.
The problem is, from my real experience, is that if I use Astra model To edit my videos, and let's say I'm making shorts like this one, With the level of effort At the minimum, so I cannot really even decrease it lower because otherwise, I need to use not Astra model, but the previous version, Sol, Terra, or Luna, which is not that great for editing videos. So if I'm using the lowest version from Astra, it looks approximately like that for each of the videos I consume from between 5–10% of my weekly consumption of tokens.
So you can imagine that, especially if you start for the first videos, I think I used, like, 20 or even 25% of the consumption of weekly, tokens. That means that, in a week, I can actually make around 10 Shorts like that, which is more than one Shorts per day. But I'm not using it only for video creation. Use it also for other things. And as a result, I think that it works, but that's actually not enough. So that's my conclusion. Astra is amazing. You should pay, I think, €100 for these models.
At the same time, it all depends on which kind of things you are doing with these models. And in terms of videos, 10 shorts will be done. That's amazing. But if you want to do also coding in parallel and something else, then I would say it's not enough. And maybe you should have Claude subscription as I have as well. So when my Astra tokens are fully consumed, I switch to Claude, and I use Claude not for editing the videos for doing some other things. So Claude for €20 currently and Astra model for €100...
per month. So I'm basically consuming €120 per month just on AI. It's very interesting To look at myself because in the beginning of the year or even maybe a couple of months ago, I was for sure not ready to do it. But now I just think that I really cannot do otherwise or I would need to spend so much more time on different elements of my life, private life, work life, that this kind of consumption of tokens, for me personally, makes sense. What about you?
