StackBlitz's browser-based AI IDE for generating and running full-stack apps instantly
TL;DR: Bolt.new carries a Risk Score of 33/100 (Mixed Signals) based on 2 documented complaints — mostly minor issues, but worth knowing about.
Bolt.new is an AI-powered development environment from StackBlitz that runs Node.js directly in the browser. It generates full-stack applications from prompts and lets you edit, run, and deploy without a local dev environment. Token-based pricing starts at $20/month.
Website: bolt.new · Category: Vibe Coding & AI IDEs · Last scanned: 20 days ago
Bolt.new's technical underpinning is genuinely clever — running a real Node.js environment in the browser via WebContainers means it can actually execute and test code rather than just generate it. That's a meaningful advantage over tools that generate static output you then have to run elsewhere.
The early 2025 complaint record is dominated by two themes: token consumption that outpaces expectations, and the gap between code that runs in the Bolt.new sandbox versus code that behaves consistently in production. The AI-generated code is often functional but not production-quality — users who don't know enough to review it carefully describe deploying broken apps. The token model also means that fixing bugs introduced by the AI itself costs additional tokens, which users find frustrating.
Help others know they're not alone. Your report is anonymous.
Each complaint type is weighted differently in the Risk Score. Billing and marketing deception weigh heaviest.
Complaints are sourced from public platforms spanning US, UK, and global consumers. Each report links back to its original source.
| Platform | Reports | Who's reporting |
|---|---|---|
| 2 | US & global users |
Documented pricing complaints, billing issues, and support failures — newest first.
Our AI scanner searches Reddit, Trustpilot, BBB, and news sources for fresh complaints from the past year, paraphrases what it finds, and adds anything new to this page. Takes up to 90 seconds.
Bolt.new bills by token consumption, which includes both the code it generates and context it reads. Users doing iterative development — ask, test, fix, repeat — describe running out of tokens within days on the base plan, including when fixing bugs the AI itself introduced.
"The AI broke my auth flow on edit 4. Fixing it cost more tokens. Paying for the AI to fix its own mistakes is not what I signed up for."
Code that runs correctly in Bolt's WebContainer sandbox doesn't always behave identically in production environments. Environment variable handling, dependency resolution, and file system differences create surprises when users deploy the generated code.
We've documented 1 billing complaint against Bolt.new — a signal worth weighing before committing to a paid plan. Its Risk Score of 33/100 puts it in the "Mixed Signals" band. See the full complaints breakdown → before deciding.
Cancellation difficulty is one of the top SaaS frustration patterns. Check the complaints page → — we tag cancel-related issues under "Billing" and "Contract Trap" categories. If none are documented yet, run a scan to surface what's currently out there.
Hidden-fee complaints fall under our "Billing Issue" category. We've documented 1 billing complaint for Bolt.new so far. See all complaints → for the full picture.
We track other vibe coding & ai ides tools and rank them by Risk Score. See our alternatives comparison → to find lower-risk options in the same category.
The highest-severity documented complaints involve billing problems. Read all 2 documented complaints on the complaints page →
Probably not in the strict legal sense — most SaaS products with bad reputations are real companies delivering a real (if disappointing) product. But "is it a scam" is the question people ask when they feel they were misled. Read our full scam analysis →
We weight each warning by severity (Low to Critical) and category, then aggregate. Lawsuits and misleading-marketing claims weigh heaviest. The current 33/100 score puts Bolt.new in the "Mixed Signals" band. Full methodology →
Each warning is paraphrased from a public source — BBB filings, Trustpilot or G2 reviews, Reddit threads, Capterra ratings, court records, or news articles. The source URL is attached to every warning so you can verify it yourself. More on our methodology →
Sibling products in the same category, ranked by Risk Score (lowest first).
AWS's AI coding assistant, formerly CodeWhisperer, deeply integrated with AWS services
Privacy-focused AI code completion with on-premise deployment option
AI code editor from Codeium with an agentic 'Cascade' workflow
Vercel's AI UI code generator that produces shadcn/ui and Tailwind components from prompts
Microsoft-backed AI coding assistant built into VS Code and JetBrains
AI full-stack app builder that generates and deploys React apps from a chat prompt