v0

Vercel's generator, for people whose site is going to end up in Next.js anyway

v0 is Vercel's AI agent for generating web interfaces and full-stack applications from natural-language prompts, producing Next.js, Tailwind, and shadcn/ui code that can be previewed live, edited visually in Design Mode, synced to GitHub as pull requests, and deployed to Vercel in one click, sold on credit-based per-user plans with a free tier and Plus at $30 per user per month.

Visit website

Overview

v0 is the AI generator built by the company that makes Next.js, and that lineage explains almost everything about it. Vercel launched v0 as a research preview in September 2023, initially as a generative UI tool that turned a prompt into a React component. It has since grown into an agent that builds and deploys whole applications, but the centre of gravity has not moved: the output is Next.js with Tailwind and shadcn/ui, and the deployment target is Vercel. If your engineering team already works that way, v0 slots into an existing workflow rather than creating a parallel one. If they do not, you are adopting somebody else's stack along with the tool.

The job it does sits in the AI generator group rather than the website builder group, and it is worth being precise about which part of that group. Lovable and Bolt aim at producing a finished application for someone who may not read code. v0 has always been closest to the developer, generating components and pages that a person is expected to review, refine, and merge. Agent Mode extended that with autonomous behaviour: web search, inspecting a live site, fixing its own errors, and calling external tools. The result is less a builder than a very fast colleague who writes the first draft in your house style.

Pricing is per user with credits on top, which is a different shape from Lovable's unlimited-seat model and worth modelling before you commit. Free gives $5 of monthly credits with a limit of seven messages a day, plus visual editing and GitHub sync, which is enough to form an opinion. Plus is $30 per user per month with $30 of included credits and a $2 daily login bonus. Business is $100 per user per month, adding training opt-out by default and centralized billing on vercel.com. Enterprise is quoted, and adds SAML SSO, role-based access control, and support SLAs. Extra credits can be bought and are shared across a team.

The ownership question, which is the decisive one for this whole group, has a good answer here and a caveat attached. Generated code is real Next.js you can open as a pull request, review, merge, and run anywhere Next.js runs, which is almost everywhere. The caveat is gravitational rather than contractual: one-click deployment to Vercel, Vercel's own backend primitives, and a component library maintained by the same ecosystem all pull you toward staying. Nothing stops you self-hosting a Next.js app, plenty of people do, but v0 is designed to make the Vercel path the effortless one and every other path a decision.

Best for

Product and engineering teams already building on Next.js and Vercel who want to compress the first draft of a marketing page, internal tool, or product surface from days to minutes while keeping the output inside their existing review, repository, and deployment workflow.

Not the right fit for

Non-technical small business owners; the entire product assumes a person who reads code, reviews pull requests, and knows what shadcn/ui is, and there is no support path that will fix your site for you.
Anyone who wants a content management system; there is no editorial workflow, so a marketer cannot publish a blog post without a developer or a separate headless CMS wired in.
Teams committed to a different stack; if your applications are Rails, Django, Laravel, or plain WordPress, v0's output is a foreign object that someone has to maintain in isolation.
Budget-sensitive solo operators who need one simple site; $30 per user per month plus credits is expensive against Carrd at $19 a year or a static site on free hosting, and buys capability you will not use.
Buyers whose priority is conversion optimization; there is no A/B testing, no traffic splitting, and no conversion analytics anywhere in the product.

How it works

1
You describe what you want, or upload a wireframe, mockup, or screenshot. v0 generates a Next.js interface using Tailwind for styling and shadcn/ui for accessible component primitives, and renders it live beside the conversation with progress indicators showing what the agent is doing.
2
You refine either by talking or by clicking. Design Mode lets you select elements on the rendered preview and adjust them visually, which is the bridge between a chat interface and the direct manipulation designers expect, and it is available even on the free tier.
3
Agent Mode handles the work a single prompt cannot. The agent can search the web, inspect a live site, detect and fix its own errors, and call external tools, which means a request like matching an existing brand or debugging a failing build can be delegated rather than manually described.
4
Backends are connected rather than avoided. v0 scaffolds data-driven applications against real backend services, so authentication, database access, and server-side logic are part of what it produces rather than something you bolt on afterwards.
5
Output goes to GitHub. Generated changes can be opened as pull requests against your repository, which means the code enters your normal review process rather than living in a vendor's workspace. That is a meaningfully different posture from tools that treat the repository as an export destination.
6
Deployment is one click to Vercel, with the platform's hosting, edge network, preview deployments, and domain handling behind it. Because the artifact is a standard Next.js application, deploying it elsewhere is possible and is a deliberate act rather than a default.

Feature breakdown

25 features in 5 modules

Generation and the agent

What separates v0 from a prompt-to-component toy.

Prompt-to-interface generation: A natural-language description produces working Next.js pages and components with Tailwind styling and shadcn/ui primitives, rendered live rather than as a mockup.
Wireframe and mockup input: High-fidelity interfaces can be generated from an uploaded wireframe, screenshot, or design file, which is far more reliable than describing a layout in prose.
Agent Mode: Autonomous behaviour including web search, live site inspection, self-correcting error fixes, and external tool calls, so multi-step work can be delegated rather than prompted one instruction at a time.
Design Mode: Click elements in the live preview and edit them visually rather than describing the change in words. Available on the free tier and the feature that makes v0 usable by designers.
Model choice on paid plans: Access to the full set of available models on Plus and above, so a task can be matched to a model rather than everything running through one default.
Real-time preview with agent feedback: The running application updates alongside progress indicators showing what the agent is doing, which makes long autonomous runs legible instead of opaque.

Code, ownership, and workflow

Output that enters your existing process rather than sitting in a vendor's workspace.

Next.js, Tailwind, and shadcn/ui output: The most mainstream React stack in current use, which means the generated code is legible to any front-end developer and hiring someone to maintain it is a normal job posting.
Pull requests to your repository: Changes can be opened as PRs for team review rather than exported as a snapshot, so AI-generated code goes through the same scrutiny as human-written code.
GitHub sync: Bidirectional synchronization with your repository, available even on the free tier, so the source is yours from the first generation.
Runs anywhere Next.js runs: The artifact is a standard Next.js application. Self-hosting or deploying to another host is entirely possible, though the product is designed to make the Vercel path the frictionless one.
Backend scaffolding: Data-driven applications are scaffolded against real backend services, so authentication and database access are generated rather than left as an exercise.

Deployment and hosting

The Vercel gravity well, which is both the convenience and the caveat.

One-click deploy to Vercel: Publishing to Vercel's edge infrastructure takes a single action with certificates, CDN, and domains handled by the platform.
Preview deployments: Vercel's per-branch preview URLs mean a generated change can be reviewed by stakeholders on a real URL before it merges.
Custom domains through Vercel: Domain connection and SSL are handled by Vercel's platform rather than by v0 itself, which is one of the ways the two products are effectively one purchase.
Core Web Vitals posture: Next.js on Vercel's edge network with server components and automatic image optimization is close to a best case for page speed in this category. The variable is what the agent generates and whether anyone audits the bundle as it grows.
SEO control through code: Because the output is Next.js, meta tags, Open Graph images, structured data, sitemaps, redirects, and robots handling are all fully controllable through the framework's own metadata APIs. There is no builder ceiling here, only whether you write it.

Teams and governance

Per-user pricing with real controls at the top.

Team collaboration with shared credits: Extra credits can be purchased and shared across a team rather than being trapped in one person's allowance.
Centralized billing on vercel.com: Plus and Business roll into Vercel's billing, so a team already paying for Vercel does not add a separate vendor relationship.
Training opt-out by default on Business: The Business tier turns off use of your data for training by default rather than requiring you to find a setting, which matters for proprietary product work.
SAML SSO and role-based access control: Enterprise adds SAML single sign-on and RBAC alongside priority performance and guaranteed support SLAs.
Data never used for training on Enterprise: The strongest data commitment sits at the Enterprise tier, which is the relevant control for regulated buyers.

Limits and daily allowances

How the meter actually behaves, which is the thing people misjudge.

Free tier daily message cap: The free plan includes $5 of monthly credits and caps usage at seven messages per day, so evaluation is real but sustained work is not.
Daily login bonus on paid plans: Plus and Business add a $2 daily credit bonus on top of the $30 monthly allowance, which rewards regular use rather than bursts.
Credit top-ups: Additional credits can be bought at any time and are pooled across the team, so a heavy sprint does not require a plan change.
Per-user structure: Unlike credit-only competitors, v0 charges per user, so a five-person team on Plus is $150 per month before any additional credits. Model this before comparing headline prices.

Use cases

4 documented

Product team already on Vercel

A new feature needs a marketing page and an onboarding flow, and the front-end engineer is booked for three weeks.

v0 generates both in the team's existing Next.js and shadcn/ui conventions, opens them as pull requests for normal review, and deploys through the same Vercel pipeline everything else uses.

Designer who wants to hand over real code

A designer produces high-fidelity Figma mockups that engineers then reinterpret, losing detail and a week each time.

The mockup goes into v0, Design Mode handles the visual refinement, and what reaches engineering is reviewable Next.js in the right component library rather than a picture to be reimplemented.

Founder prototyping before hiring

A pre-hire founder needs a credible product surface to show investors and design partners, and cannot yet justify an engineer.

Agent Mode builds a working full-stack prototype deployed on Vercel, and the repository becomes the first engineer's starting point rather than something they have to throw away.

Engineering team standardizing internal tools

Every team builds small internal dashboards differently, and none of them are maintained after the person who built them leaves.

v0 generates them consistently in the company's standard stack, they land in the monorepo through pull requests, and Business-tier training opt-out keeps proprietary code out of model training.

Pricing

from $0 (free tier), then $30 per user per month (Plus)

Per-user subscription with an included monthly credit allowance, a daily login bonus, and purchasable top-up credits shared across the team.

Plan	Price	Includes
Free	$0 per month	$5 of monthly credits Limit of 7 messages per day Deploy apps to Vercel Design Mode visual editing GitHub sync Enough to form a real opinion. The seven-message daily cap, not the credit balance, is what stops you doing actual work on it.
Plus	$30 per user per month	$30 of included monthly credits $2 daily login bonus Access to all available models Team collaboration with shared purchased credits Centralized billing on vercel.com The tier most teams buy. Remember it is per user, so a five-person team is $150 a month before top-ups.
Business	$100 per user per month	Everything in Plus Training opt-out enabled by default $30 of included monthly credits plus daily bonus Team collaboration and centralized billing Higher-priority handling The jump from $30 to $100 buys data-handling posture more than raw capability, so it is a compliance purchase.
Enterprise	Custom quoted by sales	Data never used for training SAML SSO Role-based access control Priority performance Guaranteed support SLAs

Billing notes

Pricing is per user, not per workspace. This is the key structural difference from Lovable, which charges only for credits and allows unlimited team members.
The $30 included on Plus is a credit allowance, not a usage guarantee; heavy agent runs consume it faster than short component generations.
The $2 daily login bonus rewards steady daily use and is worth up to about $60 a month on top of the included allowance for someone who logs in every day.
Additional credits are purchasable and pooled across a team, so a sprint does not require upgrading everyone's plan.
Business at $100 per user is largely a data-governance tier; the meaningful difference from Plus is training opt-out by default rather than extra generation capacity.
Vercel hosting is billed separately from v0. The deployed application's bandwidth, function invocations, and image optimization all sit on the Vercel bill.

Value assessment: For a team already paying for Vercel and building in Next.js, Plus at $30 per user is straightforwardly good value, because the generated code enters an existing pipeline and saves reviewable engineering hours rather than producing a parallel artifact somebody has to babysit. For everyone else the value is harder to defend. Per-user pricing means a small team pays several times what Lovable charges for comparable generation, the Vercel hosting bill sits on top, and a solo operator wanting one marketing site is spending $360 a year plus hosting for what Carrd does for $19. The rule of thumb is simple: if the sentence 'we deploy to Vercel' is already true, v0 is cheap; if it is not, you are buying a stack as well as a tool.

Strengths & limitations

Strengths

The output is genuinely mainstream Next.js, Tailwind, and shadcn/ui, which makes it legible to any front-end developer and trivial to hire for.
Pull requests rather than exports. AI-generated code enters your normal review process, which is the correct posture and one most competitors do not offer.
Design Mode gives designers direct manipulation on the live preview instead of forcing every change through a chat prompt, and it is available on the free tier.
Agent Mode's web search, site inspection, and self-correcting error fixes turn multi-step work into delegation rather than a sequence of manual prompts.
Next.js on Vercel's edge network with server components and automatic image optimization is close to a best case for Core Web Vitals in this category.
SEO control is unlimited because it is code: metadata, structured data, sitemaps, redirects, and robots handling are all yours, with no builder ceiling.
Backed by Vercel, an established infrastructure company rather than a two-year-old startup, with SAML SSO, RBAC, and support SLAs available at the top.

Limitations

Per-user pricing makes it expensive for teams compared with credit-only competitors, and the $30 monthly allowance is consumed quickly by agent runs.
It presumes the Next.js and Vercel stack. Adopting v0 on a Rails or WordPress team means maintaining a foreign artifact nobody else touches.
No CMS of any kind, so content publishing requires a developer or a separately wired headless CMS, which rules it out for content-led marketing sites.
No A/B testing, personalization, or conversion analytics, so a performance marketer gets nothing from this product directly.
Portability is real but gravitational pressure is strong: one-click Vercel deployment, Vercel's backend primitives, and Vercel's billing all make staying the path of least resistance.
Completely unsuitable for a non-technical owner. The product assumes code review as a normal activity, and there is no support path that repairs your site for you.
Hosting is a separate bill. The v0 subscription does not cover the deployed application's bandwidth, function invocations, or image optimization on Vercel.

Head-to-head comparisons

5 alternatives

v0 vs Lovable

from $0 (free tier), then $25 per month (Pro)

The direct comparison. Lovable charges $25 a month with unlimited team members and meters only credits, and ships a more complete managed backend covering auth, databases, and payments for people who want a whole application. v0 charges $30 per user, generates Next.js rather than Vite React, and integrates with your repository through pull requests rather than a synced export. Choose v0 if you are already a Vercel and Next.js shop and want AI output inside your existing review process; choose Lovable if you want the cheaper seat structure and a self-contained application.

Full v0 vs Lovable comparison

v0 vs Durable

from $0 (free plan), then $25 per month, or about $22 billed annually (Launch)

Opposite ends of the AI generator group. Durable produces a finished small-business website with invoicing and a CRM in about a minute for someone who will never see code. v0 produces reviewable Next.js for someone who insists on seeing it. There is no overlap in buyer: a landscaping company should buy Durable, and a product team should buy v0, and either buying the other would be a mistake.

Full v0 vs Durable comparison

v0 vs Framer

from $10/mo (Basic, per site)

Framer is the designer's fast path: beautiful sites with a real CMS, sophisticated animation, and no code to maintain, hosted by Framer. v0 is the developer's fast path: code you own, infinite extensibility, and no CMS at all. For a marketing site that a marketing team must maintain without engineering, Framer wins outright. For a product surface that will grow into software, v0 wins.

Full v0 vs Framer comparison

v0 vs Webflow

from $0 (Starter on a webflow.io subdomain), then about $15 per month billed annually for Basic

Webflow offers precise visual control, a genuine CMS, and clean published output, at a several-week learning curve and a stacked bill of site plans, seats, and add-ons. v0 offers no visual design system and no CMS, but produces a codebase with no ceiling. Webflow is the better answer when non-technical people must own the site day to day; v0 is better when engineers do and Webflow's constraints would eventually become the problem.

Full v0 vs Webflow comparison

v0 vs Replo

from $0 free plan, then $99 per month for Starter

Replo builds conversion-focused pages inside Shopify with commerce data and testing native to the platform. v0 knows nothing about Shopify and produces standalone Next.js. If the page needs a product catalogue, inventory, and checkout, Replo is the correct tool and v0 would mean rebuilding commerce infrastructure you already have. If the page is a separate product experience, v0 is.

Full v0 vs Replo comparison

Implementation & onboarding

Setup time: A generated, deployed page within the first hour on any plan, since v0 and Vercel share an account. Integrating it into an existing repository with pull requests takes an afternoon of getting conventions right and is where the value actually starts.
Learning curve: Low for a front-end developer, who is already fluent in everything v0 produces. Moderate for a designer, who will live in Design Mode and occasionally need help. Prohibitive for a non-technical owner, who will hit a wall the first time the agent generates something wrong.
Onboarding: Fully self-serve, with documentation covering generation, Design Mode, GitHub integration, and deployment. Enterprise adds support SLAs and priority performance; there is no assisted onboarding below that.
Migration notes: There is nothing to import; you start by describing or uploading a design. Migrating out is a normal Next.js operation: the application runs on any host that supports the framework, including self-hosted Node. The friction is not the code, it is anything you have come to rely on from Vercel's platform, such as edge functions, image optimization, or preview deployments, which have equivalents elsewhere but not identical ones. Plan that dependency deliberately rather than discovering it during a migration.

Platform, API & security

Platforms: Web application at v0.appVercel deployment platformGitHub integration with pull requestsRuns on any host supporting Next.js
API: The primary developer surface is the generated Next.js codebase in your own repository, plus Vercel's platform APIs and the AI SDK ecosystem the same company maintains; agent tool-calling extends to external services.
Compliance: SOC 2 through Vercel's compliance programGDPRTraining opt-out by default on Business, and no training on customer data at Enterprise
Data residency: Handled through Vercel's platform region configuration and enterprise agreements rather than as a v0-level setting.
SSO: SAML single sign-on and role-based access control at the Enterprise tier.
Security notes: The most relevant control is data handling: Business turns off training on your data by default and Enterprise commits that data is never used for training, which matters because prompts and generated code often contain proprietary product logic. Because output arrives as pull requests, generated code passes through your existing security review rather than going live unexamined, which is a structurally safer posture than tools that publish directly.

Support & resources

Channels: Documentation and in-product guidanceCommunity supportPriority performance and guaranteed SLAs at Enterprise
Documentation: Product documentation at v0.app covering generation, Design Mode, Agent Mode, GitHub integration, deployment, and credits, alongside Vercel's extensive Next.js documentation.
Community: Large developer community around Vercel, Next.js, and shadcn/ui, with substantial third-party tutorial coverage and an active presence on developer social platforms.

Company

Founded: 2015
Headquarters: San Francisco, California, United States
Ownership: Venture-backed (Vercel Inc.)
Founders: Guillermo Rauch
Employees: Several hundred at Vercel as of 2026
Funding: Vercel has raised substantial venture funding across multiple rounds, including a $250M Series E in 2024 at a reported $3.25B valuation, with later rounds reported at considerably higher valuations.

Funding history

Round	Amount	Year	Notes
Series D	$150M	2021	Led by GGV Capital at a reported $2.5B valuation.
Series E	$250M	2024	Led by Accel at a reported $3.25B valuation, following the launch of v0.

Timeline

2015Vercel is founded by Guillermo Rauch, initially as ZEIT, and goes on to create and maintain the Next.js framework.
2023v0 launches in September as a research preview: an AI tool that generates React components from natural-language prompts.
2024Agent Mode expands v0 beyond component generation into autonomous multi-step work including web search, site inspection, and self-correcting error fixes.
2025Design Mode adds direct visual editing on the live preview, making the tool usable by designers rather than only by developers.
2025v0 moves to its own domain at v0.app and shifts to credit-based pricing with per-user plans and shared team credit pools.
2026The product is positioned as a full-stack agent building data-driven applications, with training opt-out by default on Business and SAML SSO at Enterprise.

Integrations

GitHub with pull request creation
Vercel hosting, domains, and preview deployments
shadcn/ui component library
Tailwind CSS
Backend services scaffolded for data-driven applications
External tools via agent tool-calling
SAML SSO providers at Enterprise
Any npm package available to a Next.js project

Frequently asked questions

10 questions

What is v0?

v0 is Vercel's AI agent for building web interfaces and full-stack applications. You describe what you want or upload a wireframe, and it generates Next.js code using Tailwind and shadcn/ui, previews it live, and lets you refine it either by chatting or by editing visually in Design Mode. Output syncs to GitHub as pull requests and deploys to Vercel in one click. It is a developer tool, not a website builder.

How much does v0 cost?

The free tier gives $5 of monthly credits with a cap of seven messages per day, including Vercel deployment, Design Mode, and GitHub sync. Plus is $30 per user per month with $30 of included credits and a $2 daily login bonus. Business is $100 per user per month and adds training opt-out by default. Enterprise is quoted and adds SAML SSO, role-based access control, and support SLAs. Note that pricing is per user, and Vercel hosting is a separate bill.

Do I own the code v0 generates?

Yes. The output is standard Next.js with Tailwind and shadcn/ui, and it can be opened as pull requests against your own repository so it enters your normal review process rather than sitting in a vendor workspace. It runs on any host that supports Next.js. The honest caveat is gravitational rather than legal: one-click Vercel deployment, Vercel's backend primitives, and shared billing all make staying easier than leaving, even though leaving is entirely possible.

Is v0 code maintainable?

More so than most in this group, for two structural reasons. The stack is the most mainstream one in front-end development, so any competent developer can read it, and the pull request workflow means generated code gets reviewed before it merges rather than accumulating unexamined. The same failure mode still exists: an application grown over months of agent runs that nobody has audited will drift, and the bundle will quietly get heavier. Review as you go, not at the end.

Can v0 build a marketing site with a blog?

It can build the marketing site. It cannot give you a blog a marketer can update, because there is no CMS, no editorial roles, and no publishing workflow. The normal answer is to wire a headless CMS into the generated Next.js application, which is straightforward for a developer and entirely out of reach for anyone else. If the site's job is publishing content, Framer, Webflow, or WordPress.com are the right tools.

How fast are v0 sites and do they pass Core Web Vitals?

Next.js on Vercel's edge network, with server components and automatic image optimization, is close to a best-case configuration for page speed in this category, and comfortably ahead of Wix or a plugin-heavy WordPress site. The variable is what accumulates: client-side state, unnecessary dependencies, and unoptimized media added over dozens of generations. The framework gives you every tool to fix this, which means responsibility for it is yours.

Does v0 support A/B testing?

Not natively. There is no page-variant testing, traffic splitting, or conversion reporting in the product. Because the output is code, you can integrate any testing tool or use Vercel's own edge middleware to split traffic, but that is engineering work rather than a feature. If conversion testing is the point of your purchase, Unbounce, Instapage, or Landingi are built for it and v0 is not.

How does v0 handle SEO?

Completely, because it is code. Next.js metadata APIs give you titles, descriptions, Open Graph images, canonical tags, structured data, sitemaps, and robots handling with no platform ceiling of any kind, which is more control than any hosted builder in this category offers. The catch is symmetrical: nothing is configured for you, and a generated site with no metadata will have no metadata until somebody writes it.

Should a small business without developers use v0?

No. This is the clearest case in the category. The product assumes you review pull requests, understand a component library, and can tell when the agent has produced something wrong. There is no support desk that will fix your layout, and the per-user pricing plus separate Vercel hosting bill makes it expensive for the simple sites a small business usually needs. Wix, Squarespace, Durable, or Carrd are all better homes for that money.

How does v0 compare to a developer building the same thing in Next.js?

It is the same thing, produced faster, by the same conventions, in the same repository. That is unusual and it is the strongest argument for the tool: unlike a builder, v0 does not create an artifact your developer has to work around, it creates the artifact they would have written. What it does not replace is judgement about architecture, performance budgets, and what should not be built at all. The productive framing is that v0 writes the first draft and a developer decides whether it deserves to survive.

Editorial verdict

v0 is the sharpest AI generator in this category for the narrow audience it is actually built for. If you already ship Next.js to Vercel, it slots into your repository, your review process, and your deployment pipeline, produces code your team can read, and delivers close to best-in-class page speed and total SEO control because the output is just the framework. Nothing else in this group integrates that cleanly with an existing engineering practice. The reservations are equally clear. Per-user pricing at $30 makes it several times more expensive than Lovable for a small team, Vercel hosting is a separate bill on top, there is no CMS and no testing layer, and a business without developers should not go near it. Buy it if the sentence 'we deploy to Vercel' is already true. If it is not, you are adopting a stack alongside a tool, and there are cheaper ways to get a website.

Written by the SaaSTracker editorial team. Awards, when shown, are judged against the published criteria in our methodology.