Vibe Coding Tools Compared: Plans, Prices, Hosting and Team Controls Side by Side

Vibe coding tools publish their plans, usage units, hosting models and team controls are described on dozens of separate pages that use different terms for similar things. This page lines up Bolt, Claude Code, Cursor, Lovable, Replit and v0 using the same fields for each, taken only from each provider's own pricing pages and documentation. It is for developers, team leads and buyers who have already read what vibe coding is and now need the facts to shortlist a tool. Tools are listed alphabetically; order is never a ranking, and the page gives no verdicts.

Last checked: 2026-11-15.

This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider. Plans and prices change often, so confirm against each provider's current pages before buying.

Key takeaways

• Two shapes of tool share one label. Bolt, Lovable, Replit and v0 are browser app builders that host the project; Claude Code and Cursor are coding agents that work on a repository you already control.
• Entry paid prices sit between $8 and $30 a month. List prices for the first paid individual tier range from Cursor Plus at $8 to v0 Plus at $30 per user; each provider meters usage differently (tokens, credits, usage multiples or a shared plan limit).
• Usage units are not comparable. Bolt meters tokens, Lovable and v0 meter credits, Replit meters dollar credits and Free Mode hours, Cursor states usage as multiples of its Plus plan, and Claude Code shares the Claude plan's usage limit.
• SSO is a team or enterprise feature everywhere. Bolt, Cursor, Lovable, Replit and v0 list single sign-on on a team, business or enterprise tier, and Claude Code follows the Claude plan that grants access; no page checked lists SSO on a free or individual tier.
• Model training defaults differ by plan. Several providers exclude business data from training by default but let individual users opt out or in; check the setting for the exact plan you buy.

What counts as a vibe coding tool on this page?

The inclusion rule is objective: a tool is included when its provider's own documentation describes building or changing software from natural-language instructions, and its plans could be confirmed on an official pricing page on 2026-11-15. The six tools named in searches for this comparison all meet that rule. For the concept itself, its limits and the wider tool list (including Codex and Google AI Studio), see Vibe Coding Explained. For the models that sit underneath these tools, see Coding LLMs Compared.

How each provider describes its own product, in its documentation:

• Bolt: "an AI app builder that turns a prompt into a working web or mobile app", built by StackBlitz and run in the browser with nothing to install.
• Claude Code: "an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools", available in the terminal, IDE, desktop app and browser.
• Cursor: Cursor Agent is the editor's assistant for complex coding tasks; it can work independently, run terminal commands and edit code.
• Lovable: a full-stack AI development platform for building, changing and deploying web apps from natural-language descriptions, with editable code.
• Replit: Replit Agent "sets up your project, creates applications, checks its work, and fixes problems along the way" inside a Replit project.
• v0: "an AI agent that helps anyone create real code and full-stack apps and agents", from Vercel.

Vibe coding tools at a glance

As of 2026-11-15. Tools alphabetical. Fields are identical for each tool; "Not stated" means the provider did not state the fact on the pages checked.

The split matters more than any single feature. App builders start from a prompt and give you a running app first; coding agents start from a repository and give you diffs, commits and pull requests first. The AI Coding Assistants vs Autonomous Coding Agents page explains the governance difference between those two working styles.

How much do vibe coding tools cost?

As of 2026-11-15. USD list prices per month from each provider's official pricing page, billed monthly unless stated. Taxes excluded. Tools alphabetical.

Availability notes. Bolt: yearly billing saves "up to 28%"; Teams token allowance per member is not stated on the pricing page. Claude Code: included in all paid Claude plans and shares the plan's usage pool with chat. Cursor: prices exclude taxes and third-party model usage is billed separately. Lovable: Pro and Business prices scale with the monthly credit tier; annual billing is available. Replit: the pricing page lists further monthly amounts ($215, $425, $825 and $2,000) without naming the tier they belong to. v0: additional credits can be bought and are shared across the team.

Comparing these numbers directly is unsafe because the units differ. A token budget (Bolt), a credit price per prompt (Lovable, v0), dollar credits plus capped free hours (Replit) and a usage multiple (Cursor) or shared plan limit (Claude Code) all translate into a different amount of work depending on project size and the agent mode chosen. Run the same pilot task on each shortlisted tool and record what it consumed.

Models and agent modes

As of 2026-11-15. What each provider documents about the models behind the tool and the modes a user can pick.

Recovery features are not equivalent. Cursor states that its checkpoints are local and "not permanent version control"; Replit checkpoints cover the project state; Git-based history (Claude Code, and Git sync on the builders) survives outside the tool. For anything you intend to keep, connect a Git repository on day one.

Where does the code run, and where does it go?

• Bolt: runs in the browser; every project gets a free bolt.host subdomain, and paid users can connect custom domains. A database is created automatically when needed and can be claimed and connected to Supabase. GitHub is available for version control, and existing repositories can be imported.
• Claude Code: runs locally in the terminal or IDE, or in cloud sessions in Anthropic-managed virtual machines by default; organisations can route cloud sessions to a self-hosted environment. In Anthropic-hosted sessions the repository is cloned into an isolated VM and GitHub credentials stay on Anthropic's servers, attached by a proxy.
• Cursor: the editor runs on the developer's machine; cloud agents and code hosting are listed on Premium and above.
• Lovable: apps publish to a live URL, and each project's codebase can be synced to GitHub, GitLab or Bitbucket. Lovable states that users keep full ownership of their code.
• Replit: agents build and test in Replit's cloud environment; all artefacts in a project publish together. The Enterprise page lists isolated cloud environments for agents and BYOC (bring your own cloud).
• v0: one-click deployment to Vercel infrastructure, or a pull request for review; GitHub sync and Design Mode visual editing are listed on the free plan.

The exit path is the practical test: before building anything you will keep, confirm the code can leave the platform through Git sync, import or pull requests. Hosted builders also hold the app's secrets and database keys; the AI App Builder Security Guide covers row-level security and key handling for that model.

Team, identity and admin controls

As of 2026-11-15. Controls as listed on each provider's pricing, enterprise or security page, by the tier where the provider lists them.

Claude Code identity and admin controls follow the Claude plan that grants access, so check the Claude Team or Enterprise plan features rather than the Claude Code docs alone. Every one of these tools holds credentials to code hosts, databases or deployment targets, which makes the tool itself a non-human identity to inventory. See the AI Coding Tools and Secrets Sprawl page and How to audit AI coding assistant tokens and config files.

Data handling and model training

• Bolt: the pricing page does not state a training policy and points to the privacy policy; Enterprise lists data governance and retention policies.
• Claude Code: on consumer plans (Free, Pro, Max) users choose whether data is used for training; retention is 5 years when that setting is on and 30 days when it is off. On commercial terms (Team, Enterprise, API) Anthropic does not train on code or prompts unless the customer opts in, and standard retention is 30 days. Local transcripts are kept in plaintext under ~/.claude/projects/ for 30 days by default (cleanupPeriodDays).
• Cursor: "When enabled, we will not train on your data" (Privacy Mode); admins can enable Privacy Mode for a team, and new members inherit it. The security page gives no retention periods.
• Lovable: Free and Pro accounts can opt out per account; on Business, "Workspace data excluded from AI model training by default".
• Replit: the Enterprise page states "Your knowledge stays yours"; a training policy answer was not shown on the page checked. Enterprise lists advanced privacy controls.
• v0: training opt-out is enabled by default on Business; on Enterprise the pricing page states data is never used for training.

For an organisation, the practical rule is to buy the tier where exclusion from training is the default and enforce it centrally, not to rely on each developer's account setting. The Unsanctioned AI Coding Assistants page covers finding personal accounts that bypass the company tier.

How to choose a vibe coding tool: factual criteria

1. Starting point: no code yet and a need for a hosted web app points to the app builders (Bolt, Lovable, Replit, v0); an existing repository with tests and CI points to the coding agents (Claude Code, Cursor).
2. Stack fit: v0 names Next.js, Tailwind and shadcn/ui and deploys to Vercel; Bolt names JavaScript-based frameworks and Expo for mobile; Replit lists mobile apps, slides and video among project types; the coding agents work on whatever stack the repository uses.
3. Exit path: confirm Git sync, import or pull request support before building anything you intend to keep (all six document a Git route, in different forms).
4. Usage model: estimate cost from a pilot task, not the headline price, because the usage units differ.
5. Identity tier: if SSO, SCIM or audit logs are mandatory, price the tier where the provider lists them, which is a team, business or enterprise tier wherever the pages checked list them.
6. Training default: pick the tier where exclusion from model training is the default for your workspace.
7. Small vs large teams: a solo builder can start on an individual tier with personal Git; a company should standardise on one or two tools, buy the managed tier and block personal accounts, as set out in the Vibe Coding Policy Template.

For evaluation questions beyond these facts, the AI Coding Assistant Buyer's Guide gives a vendor-neutral criteria table.

Security guides for each tool

Configuration depth sits in separate pages so this comparison stays factual. Independent NHI Mgmt Group guides exist for Claude Code Security Guide, Cursor Security Guide, Lovable Security Guide, Replit Security Guide and the platform v0 deploys to, Vercel Security Guide. Bolt is covered in the AI App Builder Security Guide, and the risks common to all six are in the Vibe Coding Security Guide and Top 10 Vibe Coding Security Risks.

Practitioner checklist

• Decide first whether the use case needs an app builder or a coding agent, then shortlist within that group.
• Run one identical pilot task on each shortlisted tool and record the tokens, credits or usage consumed.
• Connect a Git repository before the first build, and confirm the code builds outside the tool.
• Check the tier that lists SSO, SCIM and audit logs, and budget for it if any of the three is mandatory.
• Confirm the model training default for the exact tier purchased and enforce it at workspace level.
• Inventory the tokens and keys each tool holds (code host, database, deployment) and give each an owner and rotation date.
• Block or monitor personal accounts on the same tools once a company tier is in place.
• Re-check plan pages before renewal: prices, usage units and tier features on these pages change frequently.

Standards and references

• Bolt: Introduction to Bolt
• Bolt: Pricing
• Anthropic: Claude Code overview
• Anthropic: Claude Code data usage
• Anthropic: Claude pricing
• Cursor: Pricing
• Cursor: Agent overview
• Cursor: Security
• Lovable: Welcome to Lovable
• Lovable: Subscription plans
• Replit: Replit Agent
• Replit: Pricing
• Replit: Enterprise
• Vercel: v0 documentation
• Vercel: v0 pricing

Related NHI Mgmt Group resources: Vibe Coding Explained · Coding LLMs Compared · AI IDE and Coding Agent Security Guide · AI App Builder Security Guide