Introduction
AI builders are easy to admire in a demo. The harder question is what happens after the first attractive screen appears. Bolt.new is a browser-first AI builder for websites, web apps, and JavaScript-based mobile projects. You describe an idea, and Bolt turns it into a working project that you can edit, preview, connect, and publish. Its documentation covers Figma and GitHub imports, framework support, database and authentication setup, Bolt Cloud hosting, domains, and browser collaboration.
Bolt.new comes from StackBlitz. Eric Simons and Albert Pai founded StackBlitz in 2017. Bolt.new followed on October 3, 2024, launching at ViteConf 24 as a product built around AI generation and StackBlitz WebContainers. So StackBlitz is the established company; Bolt is the newer product built on its browser-development technology.
We could not verify a current paid-subscriber count from a reliable source. A May 2025 report mentioned five million registered users, but that is a general-user figure, not a paid-subscriber figure. We do not blur those categories. Our team finalized its hands-on scorecard on October 1, 2026. The rankings and scores below belong to that snapshot. Current prices, model announcements, and live review pages are presented as later or separate evidence; they do not alter our original test result.
Performance
Product and model context
Bolt.new is the product experience. The models are underneath it. Bolt’s documentation describes Standard and Max as AI agents powered by large language models. Bolt selects the underlying model behind the scenes. Standard is intended for everyday building and UI updates. Max is designed for larger codebases, complex dependencies, refactoring, and open-ended tasks.
Free users receive Standard only. Bolt Forge entered the picture on September 14, 2026, as an experimental research-preview agent. The official Forge announcement names GLM 5.3 Flash, GLM 5.3, Kimi K3, and DeepSeek v4 Pro as its open-model lineup. Individual Pro plans received up to 50 times more Forge usage at no additional cost through October 14, 2026.
That offer came with a clear trade-off. Forge use was opt-in, and opted-in sessions could be de-identified and used in datasets for training open models. Teams and Enterprise workspaces were outside the preview. Bolt Lite was the smaller, limited-access experiment. Beginning September 14, it was offered at $9 per month through a waitlist or invitation. The official Lite page describes it as Forge-only, with sign-ups closing October 14. Lite is inexpensive, but its underlying models are experimental.
Where Bolt earns its keep
Bolt’s appeal is easy to understand: it puts design, code, preview, and deployment in one browser-based workspace. A team can import a Figma frame, bring in a GitHub repository, connect a database, preview the result, and publish the project without setting up a local development environment. Bolt also documents visual editing, design-system support, GitHub collaboration, and custom domains on paid plans.
That combination makes Bolt useful for landing pages, browser demos, client prototypes, internal tools, and early SaaS interfaces. A JavaScript team can use it to create a first working version, then move the code into a more controlled repository and development workflow. The first draft, however, is not the whole story.
In our scorecard, Bolt’s visual result was stronger than its Speed and Prompt Accuracy results. The independent TechRadar review tested Bolt with a cryptocurrency-calculator prompt. The build took a little over five minutes, while some competing tools completed comparable work in under 90 seconds. Bolt also added currencies that the prompt had not requested. This is why our lower scores matter.
Bolt may create a good-looking interface, but buyers still need to check responsive behavior, loading states, empty states, accessibility semantics, authentication, and error recovery. Visual polish gets attention. Usability earns trust.
The published UI-Bench paper compares ten text-to-app tools across 30 prompts and more than 4,000 expert judgments. Our research notes place Bolt fifth on that leaderboard. The result is consistent with our own view: Bolt can produce strong visual work when the brief is tightly constrained, but it does not lead the field overall. The separate human benchmark examined 288 generated applications from Replit, Bolt, and Firebase Studio. It used 205 participants and 1,071 quality-filtered comparisons. Bolt performed competitively on visual appeal but trailed Firebase on usability and trust. That distinction is important. A prototype can look finished while still asking too much of the person using it.
Our seven evaluation factors
We ranked the tools using the seven factors in our scorecard:
| Evaluation Factor | What It Measures | Weight |
|---|---|---|
| AI Intelligence | Architectural reasoning; repository and design-system understanding; context retention; debugging; tool selection; multi-step planning; recovery after failed approaches. | 15% |
| Speed | Time and consistency from request to an accepted runnable result, including latency, retries, tool efficiency, tests, and revision time. | 10% |
| UI/UX and Web-Design Quality | Visual hierarchy; typography; spacing; color; responsive behavior; interaction states; accessibility/semantics; design-system fidelity; information architecture; originality and polish. | 25% |
| Prompt Accuracy | Requirement coverage; literal adherence; scope discipline; handling of ambiguity; preservation of existing behavior; quality of follow-up revisions. | 20% |
| User Rating | Weighted public satisfaction from credible review platforms | 10% |
| Review Confidence | Authenticity/verification; sample size; relevance to the current product/version; recency; source diversity; agreement or divergence between platforms. | 5% |
| Features & Usability | Onboarding; visual editing; code/repository control; design-system and Figma import; backend/integrations; testing; deployment; collaboration; accessibility; plan limits and portability. | 15% |
We gave the greatest combined weight to UI/UX quality and Prompt Accuracy because the buyer is not merely purchasing generated code. The buyer is purchasing a path toward a usable web experience.
Comparisons Before You Buy
Overall scorecard
| Product | Overall Score | AI Intelligence | Speed | UI/UX & Web Design Quality | Prompt Accuracy | User Rating | Review Confidence | Features & Usability | Best For |
|---|---|---|---|---|---|---|---|---|---|
| v0 by Vercel | 8.95 | 9.2 | 9.2 | 8.7 | 9.2 | 9.8 | 6.0 | 9.0 | Frontend teams shipping polished Next.js products |
| Lovable | 8.79 | 8.8 | 8.4 | 9.0 | 8.5 | 9.4 | 7.5 | 9.1 | Teams building polished managed-backend SaaS products |
| Figma Make | 8.55 | 7.6 | 7.8 | 9.6 | 8.8 | 9.5* | 4.5 | 8.6 | Figma-driven product prototyping teams |
| Replit Agent | 7.96 | 8.7 | 7.6 | 7.2 | 7.4 | 9.2 | 7.5 | 8.8 | Engineers, students, and teams that want a browser-based repository |
| Base44 | 7.91 | 7.7 | 8.3 | 7.7 | 7.8 | 8.8 | 6.0 | 8.4 | A solo founder, small business, or internal team building a workflow application |
| Bolt.New | 7.65 | 7.8 | 6.5 | 7.8 | 7.0 | 8.8 | 6.5 | 8.5 | JavaScript teams building fast hosted prototypes |
*Figma Make’s asterisk is reproduced from the supplied scorecard. We did not infer a different calculation.
Reading Bolt’s results
AI Intelligence – 7.8/10. Bolt is behind v0, Lovable, and Replit Agent, but ahead of Figma Make and Base44. That places it in the capable middle: suitable for structured JavaScript work, but less convincing when the task demands deep architectural reasoning, context retention, debugging, or recovery from a failed approach.
Speed – 6.5/10. This is Bolt’s weakest score and the lowest Speed score in the comparison. It does not say every build will be slow. It says the path from request to accepted result is less predictable than it is with the alternatives. TechRadar’s five-minute-plus cryptocurrency-calculator test illustrates the point.
UI/UX and web-design quality – 7.8/10. Bolt falls behind Figma Make, Lovable, and v0, but scores above Replit Agent and Base44. Give it a detailed brief covering layout, components, spacing, states, and responsive behavior, and its chances improve. Leave the brief vague, and the output has more room to wander.
Prompt Accuracy – 7.0/10. Bolt scores last here. The practical danger is scope drift: extra content, extra functionality, or a rewrite that was never requested. That is inconvenient in a personal experiment and expensive in a client project with an approved information architecture.
User Rating – 8.8/10. Bolt scores below v0, Figma Make, Lovable, and Replit Agent, and ties Base44. This is public-satisfaction evidence. It is not a direct test of interface quality.
Review Confidence – 6.5/10. Bolt has a reasonable amount of public evidence, but the sources disagree. G2 is moderately positive. Trustpilot is much more negative. That disagreement is itself information.
Features & Usability – 8.5/10. This is Bolt’s strongest relative result. Figma and GitHub workflows, browser execution, visual edits, databases, hosting, domains, integrations, and deployment give it real practical value. Those capabilities still sit beside token limits, browser requirements, plan restrictions, and the need to inspect generated code.
The alternatives, in plain English
v0 by Vercel takes first place at 8.95, with 9.2 scores for AI Intelligence, Speed, and Prompt Accuracy. It is the better starting point for a frontend team that wants polished React or Next.js work and close adherence to a defined brief. Its current Product Hunt page shows 4.9/5 from 60 reviews, though that remains a community rating rather than a controlled design benchmark.
Lovable comes second. It leads the comparison on Features & Usability at 9.1 and also earns 9.0 for UI/UX. It is worth considering when a team wants a polished SaaS workflow with a managed backend. Figma Make owns the strongest UI/UX score at 9.6 and records 8.8 for Prompt Accuracy. It is the obvious candidate for teams whose work begins in Figma. Its lower Review Confidence score means the design evidence is stronger than the public-review evidence.
Replit Agent pairs strong AI Intelligence and Features & Usability scores with a lower UI/UX score. It makes more sense when repository access, coding control, and a browser-based development environment matter more than first-pass visual polish. Base44 does well on Speed and fits workflow applications and internal tools. It may be a better match for a buyer who wants fast business-process scaffolding rather than JavaScript repository control.
Bolt’s sixth-place position is not a declaration that it is unusable. It is a description of fit. Bolt’s strengths are browser-based JavaScript building, imports, hosted deployment, and a broad feature set. Change the weighting to emphasize integrations or deployment, and the ranking could change. Another researcher could reasonably choose a different winner, including Claude-based tooling.
Customer Review
Our October 1 scorecard used this public-rating snapshot for the six evaluated tools:
| Product | Source used | Public Rating | Votes & Reviews |
|---|---|---|---|
| v0 by Vercel | Product Hunt | 4.9/5 | 60+ reviews |
| Lovable | G2 | 4.6/5 | 433+ reviews |
| Figma Make | Trustradius | 4.4/5 | 1.7k+ reviews |
| Replit Agent | G2 | 4.4/5 | 413+ reviews |
| Base44 | Product Hunt | 4.4/5 | 41+ reviews |
| Bolt.New | G2 | 4.3/5 | 96+ reviews |
These numbers belong to a dated research snapshot. They are not permanent labels. The current Product Hunt pages, for example, show v0 at 4.9/5 from 60 reviews and Bolt.new at 4.4/5 from 47 reviews.
We also collected a wider Bolt-specific review sample:
| Platform | Rating in our snapshot | Review count | How we interpret it |
|---|---|---|---|
| Trustpilot | 1.5/5 TrustScore | 203 | Serious service-risk signal |
| Gartner Peer Insights | 4.0/5 | 1 rating | Too small for a meaningful conclusion |
| App Builder Index | 4.2 community rating; 3.9 editorial score | 79 ratings; 5 written reviews | Useful directional evidence, but mixed measures |
| G2 | 4.3/5 | 96 product reviews | Primary product-rating source |
| Product Hunt | 4.4/5 | 47 reviews | Launch and community sentiment |
| TrustRadius | 9.3/10, approximately 4.65/5 | 3 reviews/ratings | Positive but very small B2B sample |
| SourceForge | 5.0/5 | 1 review | Anecdotal |
We selected G2’s 4.3/5 as Bolt’s primary public rating because it was product-specific, B2B-oriented, and had the largest meaningful verified-review sample in our snapshot. G2 says published reviews go through moderation, reviewer-login checks, and conflict-of-interest screening. Trustpilot’s 1.5/5 is the uncomfortable part of the picture, especially for support, billing, tokens, and production use. The current Trustpilot page displays a 1.5 TrustScore, a very high share of one-star reviews, and a warning that Bolt has not recently invited reviews. That warning means the sample may not represent every customer. It does not make the complaints irrelevant.
We do not blend these platforms into one average. They are looking at different things. Product Hunt reflects launch and community sentiment. G2 is more useful for product-specific B2B feedback. Trustpilot is more sensitive to service, account, subscription, token, and billing problems. TrustRadius itself warns that simple averages can hide differences in review date, source, and rating type.
The praise is familiar: Bolt can turn a prompt into a working browser project, and its interface, integrations, and deployment workflow are attractive for prototypes and early applications. The complaints are familiar too: token consumption, error loops, support responsiveness, billing, and reliability as projects become larger. Product Hunt’s current summary captures both sides, mentioning Bolt’s browser-based prototyping strengths alongside complaints about token costs, support, and larger-project reliability. That mixture agrees with our scorecard. Bolt’s features and visual output can be good. Speed, prompt discipline, and service confidence are less dependable. This does not mean every project will fail. It means production buyers should understand the downside before they commit important work.
Pricing
We verified the official pricing information on October 3, 2026. The current pricing page shows monthly and yearly billing, advertises savings of up to 28% with yearly billing, and confirms that Teams pricing is per member with separate token allocations for each member.
| Plan | Monthly price | Annual price | Main allowance and features |
|---|---|---|---|
| Free | $0 | $0 | 1M tokens/month, 300K daily limit, public and private projects, hosting, unlimited databases, Bolt branding, 10MB uploads, and up to 333K web requests |
| Bolt Lite | $9/month | Not listed | Invite-only research preview, Forge only, marketed as up to 50× more usage than Standard mode; sign-ups close October 14 |
| Pro | $25/month | $216/year, equal to $18/month | Starts at 10M tokens/month, no daily limit, no branding, private sharing, 100MB uploads, hosting, up to 1M web requests, rollover, custom domain, SEO tools, expanded database capacity, provider choice, and AI image editing |
| Teams | $30/member/month | $324/member/year, equal to $27/member/month | Everything in Pro plus centralized billing, access management, provisioning, organization sharing, private NPM registries, and design-system features |
| Enterprise | Custom | Custom | Advanced security, SSO, audit logs, compliance support, dedicated account management, priority support, custom workflows, and SLAs |
The annual calculations at the entry tiers are:
- Pro monthly: $25 × 12 = $300; Pro annual: $216; savings: $84, or 28%.
- Teams monthly: $30 × 12 = $360 per member; Teams annual: $324 per member; savings: $36 per member, or 10%.
- Bolt Lite has no annual price listed. Twelve months at $9 per month would total $108, but access was limited to the research-preview rollout.
Here is the small pricing wrinkle worth noticing. The public page says “up to 28%” because the discount varies by plan and token tier. The billing guide describes annual savings generally as 10%, while its examples confirm $216 for the entry Pro annual plan and $540 for a larger Pro tier. Check the actual Yearly tab for the token option you want. Do not assume every plan receives the full 28% discount.
Free is useful for testing. Its 300K daily limit and 1M monthly allocation make repeated design revisions difficult, though. Free projects keep Bolt branding, have a 10MB upload limit, and cannot use custom domains. Unused Free usage does not roll over. Pro is the best value for a freelancer or solo builder who expects recurring work for at least a year. The $216 annual payment equals $18 per month, removes the daily token limit, starts at 10M tokens per month, supports rollover, removes branding, and adds a custom domain, private sharing, larger uploads, database-provider choice, and higher web-request capacity.
Paid tokens can roll over for one additional month, but an active paid subscription is required to use them. The break-even point arrives after roughly nine months. Eight monthly Pro payments cost $200. Nine cost $225. If you expect to use Bolt for nine months or more, the annual plan is cheaper. Monthly Pro is the better choice when the project may last only a few months, cash-flow flexibility matters, or you are still testing the fit.
Teams is for genuine collaboration. The plan adds centralized administration and organization-level sharing, but the price is per member and tokens are allocated per member rather than pooled. Five team members would cost $150 per month or $1,620 upfront for the annual equivalent, with five separate member allocations. That is less appealing if one person will do most of the building.
Lite is a narrow fit: an invited user who wants inexpensive access to experimental Forge models for side projects, classes, or weekend experiments. The opt-in data-sharing arrangement and experimental model status matter, particularly for confidential code or production work. During the September 30 preview period, individual Pro users could access Forge without waiting. That was the more practical route for immediate Forge access, but the up-to-50× offer was scheduled to end October 14.
Our primary recommendation is Pro annually at $216 per year, equal to $18 per month, for a freelancer or recurring solo builder who expects to use Bolt for at least nine to twelve months. Choose Free for testing, Lite only when invited and comfortable with experimental Forge, Teams for real multi-person collaboration, and Enterprise when governance, security, compliance, and procurement requirements justify a custom contract.
Conclusion
Bolt.new is a capable browser-first JavaScript builder. It brings together Figma and GitHub workflows, hosted deployment, databases, visual editing, integrations, and a browser-based development environment. That makes it a useful tool for demos, disposable prototypes, landing pages, and JavaScript teams that want to move from an idea to a hosted project quickly.
Our October 1 scorecard places Bolt sixth overall at 7.65/10. The reason is not mysterious. Features & Usability scored 8.5. UI/UX quality scored 7.8. Speed scored 6.5. Prompt Accuracy scored 7.0. External testing and public feedback tell a similar story: Bolt can look good and produce a working project, but it may take longer than expected, wander beyond the brief, consume tokens during repair loops, or create support and billing concerns.
For most recurring individual use, Pro annually is the strongest purchase. It offers the best balance of price and capability, with no daily token limit, rollover, private sharing, custom domains, and the lowest verified long-term entry cost. If visual polish is your first priority, compare v0, Lovable, and Figma Make. If repository control and browser-based engineering matter most, consider Replit Agent. If Bolt’s browser workflow, JavaScript flexibility, hosted deployment, and integrations are the features you value most, it remains a reasonable choice provided you budget time for review, revision, and quality control.
FAQ
Is Base44 better for building a full application than for designing a visually polished website?
Our scorecard rewards the quality of the finished result as well as the tools available to create it. UI/UX quality and Prompt Accuracy account for 45% of the total weighting. Bolt scored 7.8 and 7.0 in those categories, alongside 6.5 for Speed. Its integrations, visual editing, databases, and hosting add practical value, but weaker prompt adherence and slower completion pull its overall score down to 7.65/10.
What should I check before handing a Bolt.new website to a client?
Compare the finished website against the approved brief, then check responsive layouts, loading and empty states, accessibility semantics, authentication, and error recovery. Our 7.8/10 UI/UX score supports Bolt’s visual capabilities, while its 7.0/10 Prompt Accuracy score makes scope checking especially relevant. TechRadar’s calculator test illustrates the concern: Bolt added currencies beyond those requested. A professional handover therefore requires checking both how the website looks and whether it follows the requirements.
Why does the article use G2’s 4.3/5 rating when Trustpilot gives Bolt.new only 1.5/5?
The platforms capture different aspects of the customer experience. We selected G2’s 4.3/5 from 96 product reviews as the primary product-rating source because of its product focus, B2B relevance, and review checks. Trustpilot’s 1.5/5 from 203 reviews remains a serious signal about support, billing, tokens, and account problems. We keep those findings separate because averaging them would obscure the difference between satisfaction with the builder and satisfaction with the surrounding service.
How long must I use Bolt.new before annual Pro becomes cheaper than monthly billing?
Using the article’s October 3, 2026 pricing snapshot, annual Pro costs $216 upfront, while monthly Pro costs $25. Eight monthly payments total $200; nine total $225. Annual billing becomes cheaper once you need a ninth month. Over twelve months, it saves $84, or 28%. Monthly Pro remains the better fit for shorter projects, uncertain usage, or buyers who need payment flexibility while evaluating Bolt.
Does the $9 Bolt Lite plan provide the same AI agents as Pro?
Lite uses experimental Forge models rather than Standard or Max, so its lower price comes with a different building experience. In the article’s preview period, access required an invitation or waitlist, with sign-ups scheduled to close October 14, 2026. Forge participation also involved an opt-in arrangement allowing de-identified sessions to enter open-model training datasets. The advertised up-to-50× usage was a temporary preview offer, so buyers should consider model maturity, data handling, and access terms alongside the subscription price.
