Bolt.new
Bolt.new is StackBlitz's in-browser AI full-stack app builder. It runs a Node.js runtime inside your browser tab, installs packages, starts dev servers, and generates full-stack code from prompts—no local environment setup required.
What is Bolt.new?
Bolt.new is an AI-powered full-stack development platform from StackBlitz. It runs entirely in the browser using WebContainers, a technology that boots a containerized Node.js environment inside the browser tab. This means you can prompt Bolt to generate an application, watch it install npm packages, start a dev server, and preview the running result—all without installing Node.js, configuring a build toolchain, or opening a terminal on your machine.
Unlike tools that generate UI code for you to copy into an existing project, Bolt is designed to own the full loop: prompt, run, debug, and deploy. It supports Remix, Astro, SvelteKit, Vue, and Next.js, and integrates with Supabase for authentication and databases. The product is hosted and UI-driven; there is no public REST API for programmatic app generation. Access is metered in monthly AI tokens across Free, Pro, Teams, and Enterprise tiers.
How does Bolt.new work?
You start with a natural-language prompt describing the application you want. Bolt selects a framework (or uses the one you specify), scaffolds the project, installs dependencies through the in-browser npm client, and starts the dev server inside WebContainer. As the app runs, errors surface in real time—Bolt reads them and can propose or apply fixes in the same chat thread. You can attach images, documents, or reference code files for context, up to 5 MB per file read at prompt time. For backend needs, Bolt can wire up Supabase, configure environment variables, and set up authentication flows within the browser runtime. When the app is ready, one-click deployment publishes it to a hosted URL.
Use Cases
- Prototyping a full-stack application quickly without leaving the browser or configuring a local environment
- Building and iterating on frontend interfaces with immediate preview and error feedback
- Scaffolding projects in Remix, Astro, SvelteKit, Vue, or Next.js before exporting the codebase for further development
- Connecting a Supabase-backed application—auth, database, and API routes—through conversational prompts
- Running hackathon demos or throwaway prototypes where setup time should be near zero
Pricing
Pricing model: Freemium
| Plan | Price | Billing | Included usage | Notes |
|---|---|---|---|---|
| Free | $0 | per month | 1M monthly tokens, 300K daily token cap, public and private projects, website hosting, Standard agent only, Bolt branding on deployed sites, 10 MB file upload limit, no token rollover. | — |
| Pro | $25 | per month | 10M monthly tokens, no daily cap, token rollover (unused tokens valid for two months from the start of the billing cycle), 100 MB file upload limit, custom domains, private sharing, no Bolt branding. $25/month billed monthly or $18/month billed yearly. Higher token tiers available at $50/month (26M tokens) and $100/month (55M tokens). | — |
| Teams | $30 | per member per month | Everything in Pro, plus team access controls, centralized billing, admin tools, private NPM registries, and design system support. Tokens are allocated per member and are not shared between teammates. $30/member/month billed monthly or $27/member/month billed yearly. | — |
| Enterprise | Custom | — | Quote-based. Includes everything in Pro plus SSO, audit logs, compliance support, dedicated onboarding, custom workflows, and priority support. | — |
Common Limitations
- Token consumption unpredictability · Bolt meters usage in AI tokens, which cover prompts, code generation, file context, and debugging. Simple landing pages consume relatively few tokens, but full-stack applications with authentication, database logic, multiple pages, and iterative debugging can exhaust monthly allocations quickly. Pro includes 10M tokens, but complex builds may require higher tiers or token reloads.
- File attachment hard limit · Bolt reads each attached file up to 5 MB. You can upload larger files (up to 10 MB on Free or 100 MB on paid plans), but Bolt rejects any attachment over 5 MB at prompt submission. Attachments per message are capped at 5 files on Free and 20 files on paid plans. Bolt cannot read .doc, .docx, .xls, .xlsx, .ppt, .pptx, or .zip files.
- WebContainer runtime constraints · Bolt runs applications inside StackBlitz's WebContainer, an in-browser Node.js runtime. It cannot execute native binaries, so C/C++ compilation, Rust compilation, and system binaries like make or cmake are unavailable. Git CLI is not present; Bolt uses isomorphic-git for limited Git operations. Python is limited to the standard library with no pip package installation.
- JavaScript-only build target · Bolt builds JavaScript-based applications. Files in other languages (Python, Rust, Go, etc.) can be uploaded as reference material or converted by Bolt, but they are not executed as native runtimes within the project.
- Runtime performance ceiling · Complex, resource-intensive applications can hit the edges of what a browser-hosted runtime handles smoothly. The WebContainer environment is not equivalent to a cloud VM or local development machine for heavy workloads.
- Teams token isolation · On Teams plans, token allocations are per member and are not pooled or shared. A member who exhausts their allocation cannot draw from teammates' unused tokens.
Key Features
- In-browser Node.js runtime via StackBlitz WebContainer, with no local setup
- Full-stack app generation across Remix, Astro, SvelteKit, Vue, and Next.js
- Real-time execution and error feedback inside the browser tab
- Supabase integration for authentication, database, and backend logic
- One-click deployment with custom domain support on paid plans
Bolt.new vs. v0 and Lovable
This comparison focuses on documented execution model, framework flexibility, and pricing structure. Bolt.new, v0, and Lovable occupy overlapping but distinct positions in the prompt-to-app category; the goal is to clarify what each is built to produce and under what constraints, not to rank output quality.
| Tool | Positioning | Key differences | Best fit | Pricing note |
|---|---|---|---|---|
| Bolt.new Official site File attachment limits Pricing overview | An in-browser full-stack AI app builder that runs a real Node.js runtime via WebContainer, supporting multiple frameworks. | Bolt executes the application inside the browser tab, so errors and runtime behavior surface immediately without a separate deploy step. It supports Remix, Astro, SvelteKit, Vue, and Next.js, and handles backend logic through Supabase integration. The trade-off is the WebContainer constraint set: no native binaries, no Git CLI, Python standard library only. Pricing is metered in tokens, and complex builds can consume allocations quickly. | Developers and technical prototypers who want a full-stack environment in the browser with framework choice and immediate runtime feedback, and who can work within WebContainer's constraints. | Free: $0 with 1M tokens/month and a 300K daily cap. Pro: $25/month (or $18/month billed yearly) with 10M tokens, rollover, and no daily cap. Teams: $30/member/month with per-member token allocations. Enterprise: custom. |
| v0 Official site Pricing | Vercel's AI UI and app builder, strongest in frontend design and Next.js/React component generation. | v0 produces best-in-class frontend output using React, Next.js, Tailwind, and shadcn/ui, with design quality that is generally more polished than Bolt's. Backend logic, auth, and databases are supported but feel secondary; v0 is frontend-first in its DNA. It deploys naturally to Vercel and integrates with the Next.js ecosystem. Unlike Bolt, it does not run a full Node.js runtime inside the browser tab. | Frontend developers and teams already working in Next.js/React who want high-quality UI generation and a natural path to Vercel deployment. | Free tier available. v0 Plus is $30/user/month. Higher and Enterprise plans exist. Pricing is separate from Bolt's token model and is credit-based. |
| Lovable Official site Pricing | A prompt-to-app builder aimed at non-engineers and founders, with strong opinions on stack and deployment. | Lovable pairs the prompt-to-app loop with a fixed stack—React, Tailwind, and Supabase—plus GitHub two-way sync and hosting. It is designed so someone who does not code can ship a working product, with onboarding that is friendlier for non-developers than Bolt's. The trade-off is less flexibility: the stack is constrained, and developers may find the guardrails limiting. Bolt offers framework choice and a browser runtime; Lovable offers a more guided path to a deployed product. | Non-technical founders and operators who want to ship a working SaaS-style product with authentication and a database, and who prefer a guided experience over framework flexibility. | Free plan includes 5 build credits per day (up to 30 per month). Pro starts at $25/month. Custom domains and badge removal require a paid plan. |
Comparison checked against official product information on 2026-11-08. Features and pricing may change.
Frequently Asked Questions
Is Bolt.new really full-stack, or does it only generate frontend code?
Bolt.new is designed for full-stack applications. It runs a Node.js runtime in the browser via WebContainer, which means backend routes, API endpoints, and server-side logic execute inside the same environment. Supabase integration provides database and authentication. However, the runtime is constrained by the browser environment: it cannot execute native binaries or run Git CLI commands, so certain backend workloads (compilation, system-level tooling) are not possible.
What happens when I run out of tokens on the Free plan?
Free plan tokens reset daily (300K daily cap) and monthly (1M monthly allocation). If you hit the daily cap, you wait until the next day. If you exhaust the monthly allocation, you cannot generate additional code until the next billing cycle. Free plan tokens do not roll over. Paid plans include token rollover: unused tokens remain valid for two months from the start of the billing cycle in which they were received, with the oldest tokens spent first.
Can I use Bolt.new with a framework other than React?
Yes. Bolt.new supports Remix, Astro, SvelteKit, Vue, and Next.js, in addition to React-based stacks. It is not locked to a single framework like some competing tools. The build target is JavaScript-based applications; other languages can be uploaded as reference material but are not executed as native runtimes.
What are the technical limits of the WebContainer runtime?
WebContainer cannot run compiled native binaries, so C/C++, Rust, and system tools like make or cmake are unavailable. Git CLI is not present; Bolt uses isomorphic-git for limited operations. Python is restricted to the standard library with no pip packages. Shell commands are a limited zsh-emulating subset. These constraints matter most for projects that depend on native compilation or system-level tooling.
Does Bolt.new have an API?
Bolt.new is a hosted, UI-driven product. There is no public REST or HTTP API for programmatic app generation. The open-source bolt.diy project (a separate, community-maintained codebase) provides a self-hosted alternative, but it is not the same product as the hosted Bolt.new service.
Alternative Tools
-
Lovable Lovable is an AI full-stack app builder that turns natural-language descriptions into working web applications with frontend, backend, database, auth, and deployment. It is unusually effective at getting to a functional prototype and unusually poor at producing secure production code without human intervention; third-party scans find security issues in 99% of Lovable projects, and the managed backend creates migration friction that the code-portability narrative does not fully offset. -
Replit Replit is a browser-based IDE with an autonomous Agent 3 that builds, tests, and deploys apps from natural language, running up to 200 minutes without supervision. Core at $20/month is the real entry point; Starter is a demo tier. Effort-based billing, documented agent reliability issues, and a pattern of billing and support complaints are material risks for production use. -
Cursor Cursor is an AI-first code editor built on a VS Code fork with agent mode, Tab completions, and cloud agents. -
GitHub Copilot GitHub Copilot is an AI coding agent embedded across GitHub, IDEs, the terminal, and the web. It combines unlimited completions on paid plans with agent mode, asynchronous coding agents, and code review, but its value depends on disciplined review, awareness of IP and data-training policies, and realistic expectations about PR acceptance rates and context limits. -
Ollama Ollama is an open-source local LLM runtime and model manager that makes running open-weight models on your own hardware a single command. Local inference is free and unlimited; the costs are hardware, VRAM-gated context defaults, and a documented security surface that requires deliberate network isolation. -
OpenAI Codex OpenAI Codex is an agentic coding product bundled into ChatGPT plans, spanning CLI, IDE, desktop, and cloud. Plus at $20/month is the real entry point; Free and Go are demos. Usage is credit-metered with wide ranges, ChatGPT and Codex pools merge in December 2026, and disclosed security boundary issues remain relevant for teams evaluating it for sensitive codebases.