The link you shared is private btw, so I can’t see any of it. You can change the settings by clicking on Share > Visibility (Unlisted):
Thanks so much for sharing your feedback! This all really helps. I’ll pass it along to the team as we keep working to make the product better. We’re still early in the AI journey, so it’s natural for it to make mistakes.
Which screenshot is this?
In the meantime, feel free to get a refund on credits at Help.
Your message was generated by AI and our spam bot flagged it as spam, that’s why it was hidden. I recommend you write your messages without AI to prevent this from happening in the future.
I recommend you request a refund of credits over at Help.
The string of screenshots above was right before daily $2 hit acct. Last one was right as $2 hit acct.
My hypothesis on this is WRONG though. Because on further testing V0 didn’t fix the issue, and went back to cyclical silly errors on the $2 credits. So it was just being bad on free or paid. No discrimination 
I did this. Thank you. It was very easy. I requested an amount roughly equal to the bad performance (ignored prompts, spun tires, etc). I do thank you for this. It helps. Because bleeding money fighting the bot is exhausting and defeating. I get that it’s early still too.
If there’s 1 overall theme I can imprint on the frustration point as a user, it’s how V0 moves mountains on a fresh project. It’s fast, accurate, and will write almost an entire app with a single prompt. Then as the project progresses, I feel I have to shrink requests to line by line to wrangle it under control. Even that can take 5, 10, 20, even 50+ prompts to change a single thing. And not even a complex bug. Overall it just kills the cadence of the project over time. And these aren’t even complex codebases. Maybe 5,000 lines and effectively 4 pages, on a single device. All front end too.
If it could maintain the quality of the first 50-100 to even 200 prompts in a fresh project, all the way throughout a project, this would be the experience we want. And I’m not even sure it’s “context size” related either. Thank you for your patience with me regardless @pawlean . 
Low credit v0 request seem to be more precise and accurate with their outputs. I just purchased credits and wasted the first $10 from the low quality outputs of the v0 agent, changing unnecessary things not requested. It changed the whole layout of a component without requested.
I think you implemented a noticeable “fix” … now the 2.00 daily credit is just as stupid as a topped up account. 
I guess it’s not too stupid if we keep feeding it money… 
I’ll be exploring options for the next project, I do hope you get your act together soon. This has so much potential.
I’m experiencing the same thing. First, it makes a mistake and then uses the credits to fix it. It doesn’t actually progress at all, since I’m using Premium.
Please see my response to you in the other thread / post.
There might be various factors behind this all. And again, sometimes these things happen. No AI tool agent will be perfect with a “fix for free”. So we have to learn to work with it, and ways to work around it, while the engineering teams continue to evolve the products.
Is this feature no longer available? Is this service no longer free?
I would also like to know the answer to this
I have gone through this even right now
