Bolt.new is StackBlitz’s AI app builder. You describe an app in plain English and it generates, runs, and edits full-stack web apps entirely inside your browser, using WebContainers, an in-browser Node.js runtime. Founders, designers, and developers use it to prototype and ship real products, with one-click Netlify deploys, Supabase databases, Figma import, and mobile builds through Expo. Because the whole project lives in the tool, an outage can freeze a build mid-session. Start with the live verdict at the top of this page to see whether the problem is widespread right now.

Common reasons Bolt.new stops working
- WebContainer boot failures. When the in-browser runtime cannot start, you see an error saying the WebContainer took longer than 30 seconds to boot. The site loads, but previews are dead.
- AI generation stalls. Code generation is powered by Anthropic’s Claude models, so upstream incidents at Anthropic can make Bolt stop responding to prompts or cut off mid-response until you clear the chat.
- Codestorage outages. Projects can fail to open or save while the editor itself looks fine, a pattern seen in Bolt’s own incident history.
- Token limits and runaway loops. The AI can enter an error-fix loop that eats your monthly allowance, and free users who hit the daily token cap get locked out mid-project.
- Integration dependencies. Supabase database provisioning, Netlify deploys, and GitHub sync are separate services, and their failures surface inside Bolt as errors even when Bolt itself is healthy.
- Local browser conditions. Low memory, ad blockers, VPNs, and non-Chromium browsers can stop WebContainers from booting on your machine alone, so one person’s dead preview is often not an outage.
How to fix Bolt.new when it is not working
- Check the status at the top of this page first. If our check and user reports show a broad outage, the problem is on Bolt’s side and waiting is the fastest fix.
- Hard-refresh the editor tab. Reloading forces the WebContainer to boot again and clears many stuck previews and stalled generations.
- Free up your browser. Close heavy tabs, switch to a Chromium-based browser, and temporarily disable ad blockers or a VPN, since all of these can block the in-browser runtime.
- Try a private window or clear the cache. This rules out stale files and extension conflicts without touching your project.
- Break an AI loop deliberately. If Bolt keeps failing on the same fix and burning tokens, clear the conversation and rephrase the request as a smaller, more specific change.
- Check your token balance. If you have hit the free daily cap or drained your monthly allowance, generation stops until it resets, no outage required.
- Test the failing integration separately. If the error appears during a Supabase, Netlify, or GitHub step, that provider may be having its own incident, and retrying later usually resolves it.
For a fuller checklist that applies to any AI service, see our guide on what to do when an AI tool is down.
Is Bolt.new down, or is it just you?
Bolt blurs this line more than most AI tools, because half the product runs inside your browser. If the verdict and reports at the top of this page show a spike, the outage is real, shared, and on StackBlitz’s side. If reports are quiet but your preview will not render, suspect your local environment first: memory, extensions, a VPN, or your browser choice. Opening the same project in a fresh Chromium browser on another device is the single fastest test, since Bolt has no desktop app to fall back on.
How we check Bolt.new
We determine Bolt.new’s status from two signals rather than one. We read the official status feed StackBlitz publishes for Bolt, which tracks the editor along with dependencies like codestorage, Supabase, Netlify, and GitHub, and we also run our own independent reachability check against the service from the outside. Combining both means we catch problems even when one source is slow to update, and it keeps the verdict honest. You can read more about how this works on our methodology page.
Alternatives to Bolt.new during an outage
If Bolt.new is down and you need to keep shipping, the closest substitutes are other prompt-to-app builders. Lovable covers the same describe-and-build workflow for full-stack web apps, v0 is strong for generating polished front-end interfaces, and Replit pairs an AI agent with full cloud workspaces and hosting. Confirm your backup is healthy before migrating, since these tools lean on the same model providers and can wobble together. Syncing your Bolt project to GitHub regularly makes any switch far less painful.
When Bolt.new breaks inside your own browser
What makes Bolt different from most AI tools is that the failure can live on your machine. The entire dev environment, including the Node.js runtime, executes in your browser tab through WebContainers, so a whole class of Bolt problems never shows up on any status page. The classic symptom is the boot timeout banner, or a preview frozen on a white screen while the AI chat keeps answering, a split state users routinely mistake for a full outage.
The other distinct pattern is dependency spread. Bolt’s status page tracks project codestorage, Supabase, Netlify, GitHub, and its model backend as separate components, and each can fail on its own. That is why a project can refuse to save while generation still works, or why an app can run perfectly in Bolt’s preview yet publish to Netlify as a blank page when dependencies or environment variables do not transfer.
- WebContainer boot timeout error while bolt.new itself loads normally
- AI chat responding while the preview shows a white screen or endless spinner
- Projects failing to open or save during codestorage incidents
- Preview working but the published Netlify site rendering blank