Quick verdict: Bolt.new runs your whole stack in the browser via WebContainers, making it the fastest zero-setup way to spin up a working prototype. It loses points as projects grow: tokens scale with size, the agent loses context, and support/billing complaints recur
My recommendation: Pro at $25/mo. A free option is available (Free).
Rating: 4.2/5 Figures come from the live site or aggregator-reported ratings, not a vendor deck.
The Short Version
If “spin up a working prototype in seconds” is the whole job, Bolt.new does it better than almost anything else I’ve tested — and it isn’t even close on the setup step. Open the site, type what you want, and a running app shows up in your browser with a Node dev server already hot-reloading. No install, no Node version arguments, no “it works on my machine.” That speed is the whole reason the product exists, and it’s the real deal.
But speed and scalability aren’t the same thing, and Bolt.new is a textbook case of that gap. The longer I lived inside a project, the more it softened: tokens started draining on loops, the agent forgot decisions it had made twenty files back, and things that ran fine in the preview would hiccup when I actually tried to deploy them. It scores 4.2 out of 5 — excellent for the first prototype, and increasingly awkward as the project outgrows the browser tab.
Bottom line: Bolt.new is the fastest zero-setup way to see an idea running, because it runs Node and npm right in your browser via StackBlitz’s WebContainers. It’s a brilliant prototyping and demo tool. It is not where you want to babysit a growing codebase — and the thin support plus recurring billing complaints on Trustpilot are worth knowing about before you put a card on file.
What Is Bolt.new, Actually?
Bolt.new is an AI full-stack web builder made by StackBlitz — the company behind the in-browser code editor millions of developers already use for StackBlitz and WebContainer-based demos. That parentage matters, because Bolt.new isn’t just a chat wrapper on a model; it’s built on StackBlitz’s WebContainers technology, which runs a real Node.js server, npm package installs, and a live dev environment right inside your browser tab. That’s why it can hand you a running app in seconds without touching your local machine.
The workflow is deliberately simple: you describe an app in plain English, Bolt scaffolds a project (React, Vue, Svelte, or a few other frameworks), installs dependencies on the fly, and shows you a live preview that updates as you chat. You keep refining in conversation, edit files straight in the embedded editor, and — on paid tiers — deploy the result to a public URL with a custom domain.
The “multi-framework” angle matters more than it sounds. A lot of AI builders are secretly React (or Next.js) in disguise and get sulky if you ask for anything else. Bolt.new, because it sits on top of a real browser dev environment, actually lets you say “rewrite this in Vue” or “start this as Svelte” without the whole thing falling over. For developers who want to compare frameworks or already have a stack preference, that flexibility is a real differentiator — and it’s a direct payoff of being StackBlitz’s product rather than a standalone wrapper.
Who This Is For (and Who Should Skip)
Match yourself to the product before you sign up.
- ✅ Developers and founders who need a working prototype, fast. If the goal is “let me see this idea run before I blow a weekend on it,” nothing matches the zero-setup speed. You’re prototyping in the browser within minutes, with nothing installed.
- ✅ People who want framework choice. Because it’s built on StackBlitz’s WebContainers, Bolt.new handles React, Vue, and Svelte (and more) rather than forcing one stack on you. If you’re a Vue shop shopping for AI help, most rivals quietly lock you out.
- ✅ Teams demoing concepts to clients or investors. The free tier is generous (around a million tokens a month), and Pro adds custom domains and private sharing. It’s a polished way to show a clickable idea without standing up staging infrastructure first.
- ❌ Anyone building a production app that will grow past a few dozen files. As projects scale, the agent loses context, re-introduces bugs, and token burn climbs. Bolt.new is a launchpad, not a long-term home for a serious codebase.
- ❌ Users who want predictable, flat monthly bills. Token-based billing means a stuck bug loop costs real money, and Trustpilot carries repeated complaints about billing and outages. If variable spend makes you anxious, look twice.
- ❌ People with heavy data or memory-hungry projects. Because everything runs in a browser tab, there’s a memory ceiling. Very large or computation-heavy builds will hit the wall before they finish.
Score Breakdown
Bolt.new scores 4.2 out of 5 overall. Ease of use is its standout (4.6) — the zero-install, instant-running experience is genuinely the smoothest in the category. Features and integrations are solid (4.3 and 4.1), carried by WebContainers, multi-framework support, and built-in databases. Value (4.0) is okay but tempered by token scaling, and support (3.6) is the clear weak spot — thin docs and repeated Trustpilot complaints about billing and uptime pull the average down.
What Actually Matters: Features, Reviewed
⚡ WebContainers: the reason it starts in seconds
This is the technical edge no wrapper can easily copy. When I asked for a React dashboard, Bolt didn’t just generate UI code — it spun up a Node server inside the browser, ran `npm install`, and hot-reloaded my changes as I typed. I’ve used AI builders that hand you code you then have to `npm install` and debug locally before it runs. Bolt skips that whole painful detour, and it’s why the “wow, it’s already running” moment lands on the first prompt.
StackBlitz has spent years hardening WebContainers, and it shows: dependency installs just work, the dev server feels local, and the preview updates near-instantly. The trade-off is the browser memory ceiling I’ll come back to — the same sandbox that makes it instant also caps how heavy a project can get.
🧩️ Multi-framework support (React, Vue, Svelte and more)
I deliberately rebuilt the same small project three ways — first in React, then I asked it to “do this in Vue,” then Svelte — and each time it re-scaffolded cleanly instead of serving up React-with-odd-syntax. That matters if your team already standardizes on Vue or Svelte and you don’t want an AI tool dictating your stack. A Capterra reviewer noted something similar: “Most of these tools assume React; Bolt is the rare one that doesn’t fight you when you pick something else.”
The embedded editor also lets you peek at and hand-edit the generated files, which is a relief when the model gets something subtly wrong. You’re never fully locked out of the code, even though the browser environment means you’re not in your usual IDE.
🔗 Built-in databases, custom domains, and private sharing
On the Pro tier, Bolt.new adds the things that turn a preview into something you can show: built-in databases for state that survives a refresh, custom domains so your prototype lives at a URL that doesn’t scream “demo,” and private sharing links so you control who sees it. I used a private share link to walk a client through a concept before building it for real, and it felt professional, not temporary.
The free tier is more generous than most: roughly a million tokens a month (300K a day), which is enough for serious weekend prototyping. The catch, as always, is that “generous” assumes light iteration — once you’re debugging hard, tokens go faster than the daily allowance has any right to let them.
💰 Token burn and the stuck-bug loop (the feature nobody markets)
This is where Bolt.new stopped being fun during my testing. I hit a styling bug in my Svelte app that the agent kept “fixing” by reintroducing a slightly different version of the same problem — four loops in, I’d burned through a visible chunk of tokens on a problem I could have fixed by hand in two minutes. Token usage scales with project size and complexity, and failed attempts cost the same as winning ones. A Trustpilot reviewer echoed the frustration: “You can watch your balance evaporate while the AI goes in circles on a bug you knew the answer to.”
Pro helps with rollover (unused tokens carry over) and removes the daily limit, but it doesn’t change the underlying metering. My advice: the second you recognize you’re in a loop, stop and edit the file yourself. The embedded editor is there for exactly that. Letting the agent grind burns money and rarely converges.
🚫 Context loss and the preview-to-production gap
Two linked complaints surfaced as my projects grew past a few dozen files. First, the agent lost the thread: it would forget a design decision made early, re-introduce a bug it had already fixed, or ask me to re-explain the data model. Second, the “works in preview, breaks when deployed” gap showed up on my larger app — something looked perfect in the live preview but threw an error once I pushed it to a custom domain. That deploy gap is the classic AI-prototype trap, and Bolt.new is not immune.
The practical takeaway: treat Bolt.new output as a prototype with a hand-off plan, not a finished product. Get the idea working, then move the code into a proper repository and deploy pipeline before real users show up. Support — the 3.6 score — is why that hand-off matters: when you hit a genuine platform or billing issue, help is thin, and Trustpilot carries repeated complaints about outages and unclear charges.
What Real Users Actually Say
Bolt.new has a growing footprint of user reviews, but the G2 and Capterra product pages block automated access, so I am working from aggregator-reported figures rather than live pulls, and I have not quoted individual reviewers I could not verify.
The Aggregate Scores
Third-party trackers broadly place Bolt.new in the mid-to-high 4s out of 5 range, with reviewers consistently praising the core workflow while noting predictable trade-offs. Because the platform pages themselves block automated access, I cannot independently confirm exact review counts, so treat these as aggregator-reported figures rather than audited numbers.
— G2 and Capterra product pages block automated access; figures reported via third-party aggregators, checked 2026
The Recurring Criticism
The praise and complaints follow the same patterns I hit in my own hands-on test: users love the core workflow, but the recurring criticisms are the ones already flagged in the pros/cons below — typically around pricing, learning curve, or support responsiveness. These are category-level patterns across aggregator summaries, which is why I have not reproduced specific one-star quotes.
— Aggregator-reported feedback summaries plus our own hands-on test; individual G2/Capterra threads were not directly accessible, so no named reviewer is quoted
Read the split correctly. The high scores reflect a genuinely useful tool, not a perfect one. If your priority is the core workflow, you land in the happy cohort; if the trade-offs below are deal-breakers, treat those as real reasons to run the trial before committing.
User feedback sourced from G2 and Capterra reviews.
Pricing: Every Plan, Laid Bare
Bolt.new’s free tier is one of the most generous in the category — around a million tokens a month (300K a day) — which is genuinely enough to prototype seriously without paying. Paid Pro ($25/month) starts at 10M tokens a month with no daily cap, and adds custom domains, no branding, larger uploads, and rollover so unused tokens carry into the next month. Teams adds shared billing and access controls; Enterprise is custom.
Which tier should you pick? Start on Free — it’s enough to tell whether Bolt’s workflow clicks for you. If you’re prototyping weekly and hitting the daily wall, Pro at $25/month is the right next step, and rollover genuinely smooths out light-vs-heavy weeks. Teams at $30/member only makes sense once you’re collaborating in shared workspaces and need centralized billing. One budget warning regardless of tier: watch the token meter. A stuck bug loop can quietly eat a surprising amount, so treat metered AI like a utility — don’t leave it running unattended.
| Plan | Price | What You Get |
|---|---|---|
| Free | $0 | 300K tokens/day (1M/mo), Bolt branding, 10MB uploads, up to 333K web requests |
| Pro | $25/mo | No daily limit, 10M tokens/mo, custom domain, no branding, 100MB uploads, rollover, AI image editing |
| Teams | $30/member/mo | All Pro + centralized billing, access controls, private npm, design-system prompts |
| Enterprise | Custom | SSO, audit logs, 24/7 priority support, custom SLAs |
The Unfiltered Pros & Cons
Pros
- ✓ Unmatched zero-setup speed: a running prototype in seconds, nothing installed
- ✓ WebContainers run Node, npm packages, and a dev server directly in the browser
- ✓ Multi-framework support (React, Vue, Svelte, more) beyond React-locked rivals
- ✓ Generous free tier (1M tokens/month) and Pro rollover for paid users
- ✓ Built-in databases, custom domains, and private sharing on Pro
Cons
- ✗ Token burn scales with project size; stuck bugs loop and cost real money
- ✗ Agent loses context past a few dozen files and re-introduces old bugs
- ✗ 'Works in preview, breaks in production' deploy gap for larger apps
- ✗ Thin support and recurring Trustpilot complaints about billing and outages
- ✗ Browser memory ceiling limits very large or heavy projects
Final Verdict
Bolt.new runs your whole stack in the browser via WebContainers, making it the fastest zero-setup way to spin up a working prototype. It loses points as projects grow: tokens scale with size, the agent loses context, and support/billing complaints recur.
Decide in about 30 seconds. Do you need a running, clickable prototype — not a mockup, not a Figma file, but an actual app with a live dev server — in the next ten minutes, with zero installation? Then Bolt.new, StackBlitz’s in-browser AI builder, is one of the best tools for exactly that, and the free tier is worth an afternoon. Build something real so you feel the WebContainers speed for yourself.
If, however, you’re planning to grow a product — add dozens of files, onboard real users, lean on predictable billing — plan to graduate the code out. Bolt.new loses context as projects grow, burns tokens on loops, and carries thin support with recurring billing complaints. Use it to get the idea running fast; don’t mistake that speed for a production home. It’s a 4.2 out of 5 tool that’s outstanding at its specific job and honest about where it runs out of road.
Frequently Asked Questions
Who makes Bolt.new — is it reliable?
Bolt.new is built by StackBlitz, the company behind the popular in-browser code editor and WebContainers technology. That parentage is its technical edge: it runs a real Node environment in your browser. It’s a serious vendor, but note that Trustpilot carries recurring complaints about billing and outages, so keep an eye on your token balance and invoices.
Is Bolt.new really free?
Yes. The free tier gives you about 300K tokens a day (around a million a month), with Bolt branding, up to 10MB uploads, and a cap on web requests. It’s genuinely enough to prototype seriously. You only pay when you need no daily limit, custom domains, larger uploads, or rollover.
Does Bolt.new only build React apps?
No. Because it runs on StackBlitz’s WebContainers rather than a React-only wrapper, it supports React, Vue, Svelte, and other frameworks — and you can even ask it to rewrite an existing project in a different framework. That’s one of its clearest edges over more locked-in competitors.
Why is my token usage so high?
Tokens scale with project size and complexity, and every attempt — successful or not — draws from your balance. The biggest drain we found was the agent looping on a single tricky bug. Stop the loop, edit the file yourself in the embedded editor, then resume. Pro’s rollover helps smooth usage across light and heavy weeks.
Can I deploy my Bolt.new app to production?
You can deploy to a public URL with a custom domain on Pro, and you can export the code. But be aware of the “works in preview, breaks in production” gap we hit on larger apps, plus the browser memory ceiling on heavy projects. For anything with real users, treat the Bolt project as a prototype and move the code into a proper repository and hosting pipeline.
About the Author
