I’d like to escalate feedback from the community regarding challenges following the transition from V0 Dev → V0 App. A growing number of users are expressing frustration across the forums and external review platforms:
Lower quality output and difficulty completing builds.
As someone who built strong prototypes on V0 Dev, I’ve personally struggled with V0 App. For example, a recent attempt to migrate custom auth to Clerk led to 20+ failed iterations, costing ~$10, despite clear prompting. This highlights how fragile iteration has become and why new/existing users must be guided to the V0 Handbook.
It’s clear the Dev → App transition has been a major change. Removing model access and altering workflows without sufficient onboarding or credit safeguards has understandably caused friction. Many may feel this should have been introduced through a public beta with free credits to gather feedback first.
Recommendations
To address concerns constructively, I suggest:
Transparent Communication
Publish a blog post acknowledging user frustrations, explaining what’s being improved, and clarifying credit usage.
Send a group-wide email linking to the post.
Host a video Q&A session to directly address concerns.
User Support & Retention
Provide free credits ($25–50) to all users or those affected by build issues.
Reinstate user model selection (e.g., GPT-5, V0 low/medium/high, plus others, pehraps Opus 4.1).
Improve iteration guidance by prominently linking the Handbook (example).
Critical Product Improvements
Model Selection: Allow choice between GPT-5 and tiered V0 models.
Build Mode Indicator: Show whether in Build (default) vs. Answer/Query mode. Bolt AI does this.
Bug Fixing: Users should not pay repeatedly for fixing invalid code. Consider a “Fix Code” option with free or discounted retries.
Design Mode: Make all text/image elements fully editable (see Readdy.ai as a strong example).
Client Access: Enable restricted client logins for design-only changes, similar to WordPress page builders.
Long-Term Transparency
Publish a public roadmap.
Introduce a feedback / feature voting platform such as Featurebase.
The V0 community is deeply invested in this platform and eager to use it for upcoming projects. However, the current friction risks negative adoption and reputation. I hope this feedback can be taken seriously, and I’d be glad to provide further detail if useful.
I wanted to provide some additional feedback after testing the Supabase starter template and attempting to extend it with basic task management. This templates privides SB Auth login / logout with no functionality, acting as a good starting point with Auth.
Test Case
I used the most basic example with a simple, clear Steps
“Implement a CRUD (Create, Read, Update, Delete) task management feature in the application, ensuring that tasks are stored in Supabase and are associated with the currently logged-in user. The functionality should allow users to create new tasks, view their existing tasks, update task details, and delete tasks as needed. Ensure that the user interface is intuitive and responsive, providing a seamless experience for managing tasks. Additionally, incorporate user authentication to securely link tasks to individual users.”
Issues Encountered
The generated app failed with multiple runtime errors and race conditions, preventing the task dashboard from loading.
Several “Fix Code” actions were suggested in the chat interface, but these introduced further errors.
New Tailwind warnings cluttered the console, making debugging harder.
Previews were unreliable — in some cases the deployed app worked, while the in-editor preview did not. This makes it unclear whether to keep debugging locally, adding credit usage or just deploy repeatedly.
Cost Concerns (Again)
Adding task management cost $1.23 in credits, with additional fixes pushing the total to ~$2.50 for a non-working app in preview mode.
This feels unsustainable compared to earlier V0.dev usage:
Two months ago, I built a working music collaboration app in 3 days for under $20. Iterated 90 versions, fully working app - sign up and try: MusicShare - Discover and Share Amazing Music
The same type of iteration now seems far more expensive, as the V0 Agent performs significantly more steps/compute.
Summary
Current instability (runtime errors, broken previews) makes it difficult to know whether to trust the output.
The increase in credit usage per iteration makes building even simple apps financially impractical compared to previous versions.
This highlights unequivocably that Fix Code credit usage MUST be addressed including model selection.
V0 Dev was brilliant, V0 App is a major regression.
I hope management very carefully considers both stability and cost efficiency going forward — otherwise V0 risks becoming non-viable for many users who want to build and iterate quickly. Currently I’m not looking to spend further credits until there is a significant update & improvement on these issues.
To highlight Usage Cost changes, the attached data shows how credit usage and costs have dramatically shifted since the release of the V0 App on August 12th.
Because we can only access the last month of usage, I encourage everyone to check their own data now.
Before V0 App (August usage): My highest usage was just $0.17, including messages with the briefly available GPT-5 (at $0.08–$0.10).
After V0 App (Aug 13–14): Average costs at least doubled, often 3–4× higher.
Recent tests (Aug 27–30): Costs increased even further. Many of these were simply for fixing invalid code or runtime errors — in one case, 20 attempted changes all failed, including when forking the App.
Looking ahead, I’m hopeful since the release of GPT-5 (with up to a 7× cost reduction) will enable model costs to come down. See attached list for comparison.
Key takeaway: It’s absolutely critical that users have the ability to:
Choose their model
Operate in build vs. discussion mode
Access fair and predictable bug-fixing usage
As a comparison, this alarm monitoring dashboard example was created in 2.5 hours, back in late June, costing about $5, around 30 iterations, including detailed Help screens & dark mode support: https://v0-alarm-monitoring-dashboard.vercel.app/
I have totally experienced excessive credit usage - burned through $120+ in a few days like it was nothing. Also tends to get stuck on problems without being able to solve them. When I modify the prompt to give better guidance it still does things that are stupid.
But I have also found it is able to deal with more complex code in ways it couldn’t before.
Would love to see some refunds come our way for early adapters.
To be completely honest I don’t even care about the credits anymore I just want the website to look nice. I used to be really obsessed with v0 and have built so many of my previous website with this tool however now I don’t even want to risk changing them and updating because of how horrible v0 has became. Every website just looks the same. Following the same exact structure and it is really ugly. I would save so much more time if I coded the whole website by myself at least I wouldn’t spend tons of dollars on asking “agent AI” to fix this fix that.
Another problem is that for some reason recently it doesn’t update the website it’s only changes small things like the colors and icons but it no longer changes the structure. There are no good designs and code blocks being used anymore.
Yeah, it’s such a shame, for pure website creation and basic V0 type functionality, https://readdy.ai/ is superb, very affordable, and developing fast. They are a several hundred person team in Singapore, and their designs rock.
V0 seriously need to get on the ball regarding the major V0 regression, I hope that this evenings webinar will be addressing that rather than the impending support to build mobile Apps.
I’ve just seen the Design System annoucement and I’m sure it’s all wonderful, but it in no way addresses the dozens of issues users are having now with V0 generation and I don’t see any need for it right now.
Many users are seriousy upset, including myself, and are having serious issues with V0.App credit usage & code quality. It needs to work as it did before V0.App - this is a really really serious problem that is going to tank the V0 reputation if not addresses asap.
Can the team PLEASE PLEASE PLEASE PLEASE PLEASE PLEASE PLEASE address this situation, before it becomes existential. This session should have been how all the complaints are to be addressed.
Today’s v0 Styles announcement was about one feature that many people have requested. Styles can be used to create consistent UIs and save you from needing to spend credits to simply modify colors.
@alexcumbers-9492 I appreciate your concern about the success of v0. Work is always being done to iterate on code quality and user experience. Tiny changes are deployed every day to address feedback, we just don’t make a big announcement about most of them.
If there’s something specific you’d like me to share with the team, I’d be happy to pass it along.
Hi, I’m afraid you’re really not understanding how things should work or adressing ANY of the dozens of serious user concerns. Obviously it’s great that small fixes are released to address issues, and these should be detailed in the changelog.
What I & MANY others are saying is SERIOUS, you need a detailed blog post addressing ALL the forum concerns raised, and how the issuses are to be fixed, otherwise people will leave V0 for other platforms & you’ll suffer reputatonal damage, as evidenced by 80% 1* ratings on TP: https://www.trustpilot.com/review/v0.dev
My & others specific concerns are detailed extensively, which took considerable time to compile. There are serious competitors in this space such as Readdy AI who are advancing rapidly.
It’s up to you, Please do something before users leave the platform.
@anthony-shew Hi, I’m sure this is a great addition, but can the team please address the very obvious dozens of users calling out very serious regressions experienced is failed App generation and excess credit usage ASAP. Thanks
