Framer is a no-code website builder whose 2026 identity is agent-driven: agents design on the canvas, run the CMS and write components, with hosting, CMS and SEO built in. Plans run from free to $30 a month, plus per-seat and credit charges.
Used this tool? Rate it
Used this tool? Rate it
Framer is a no-code website builder that handles the whole path from blank canvas to a live, hosted site. You design visually rather than writing markup, and the platform serves the result on its own infrastructure. Where it differs from most builders is its origin: this was a designer's tool first, and it still behaves like one, favouring direct manipulation and precise control over the template-filling approach common to the category.
Its identity in 2026 has shifted decisively, and any honest description has to say so. The homepage now calls it the design agent for every step from idea to launch, and four agent surfaces anchor the product: designing on the canvas, running the CMS, writing code, and connecting to outside AI. This is not AI bolted onto a website builder as a side feature. The pricing model has been rebuilt around AI credits, the navigation carries dedicated AI and Agents entries, and the marketing leads with the agent rather than the editor. At the same time, the site is careful to say that agents work alongside you, not instead of you, which is a more restrained claim than much of the category makes.
The underlying product, though, remains a website platform, and that is what you are actually buying. Hosting, security, analytics, CMS and SEO all ship built in rather than assembled from plugins, and the pricing is metered in the units a hosting product uses: bandwidth, CMS collections, custom domains, editor seats. Understanding it as a website platform that has grown a capable agent layer — rather than as an AI tool that happens to output websites — is the framing that makes its costs and limits make sense.
Some corporate context explains the scale of the operation. TechCrunch reported a $100 million Series D in August 2025 at a $2 billion valuation, led by Meritech, for a company founded in Amsterdam by Koen Bok and Jorn van Dijk, who sold their design studio to Facebook in 2011. The same report cited $50 million in ARR and more than 500,000 monthly active users, with Miro, Perplexity and Scale AI among named customers — figures that originate with the company rather than an independent audit. Its chief executive framed the round as doubling down on enterprise growth and AI, which is precisely what the 2026 product reflects.
There is a free tier that is genuinely useful for evaluation, and a paid structure that is more layered than the headline numbers suggest. Four tiers are published openly: Free at zero, Basic at $10 a month, Pro at $30 a month and a custom Enterprise plan, with yearly billing offered alongside monthly. The free tier gives you 500 credits to try, a Framer subdomain and 1 GB of bandwidth. Basic adds a custom domain, two CMS collections, 50 GB of bandwidth and 1,000 credits a month. Pro raises that to ten CMS collections, 100 GB of bandwidth and 3,000 credits a month, and includes the agent features and branching. Enterprise brings custom limits, unlimited editors, SCIM and SSO.
The part that catches people out is that the subscription is only one of four cost inputs. Seats are billed separately at $20 per editor and $10 per content editor, so a five-person team costs substantially more than the tier price implies. Credit allowances step up within a tier, reaching 15,000 a month on Basic and 100,000 on Pro, which tells you the vendor expects heavy agent users to buy well beyond the included amount. And localization is an add-on rather than an included feature, and it draws on the same credit pool as the agents. Anyone budgeting for this should price all four axes together — tier, seats, credits and add-ons — because for a working team the subscription line is often the smallest of them.
Consumer review scores are poor, and the detail matters. Its Trustpilot score is 1.6 out of 5 across 135 reviews, with 76 percent of them one star. Crucially, 75 of those reviews were left in the last twelve months, so this is not a stale backlog. The recurring themes are commercial rather than creative: reviewers report that simple things turn complicated fast, pricing jumps, and you hit limits way sooner than expected, and that the learning curve is steep and support leans on AI bots, with email replies taking up to 24 hours.
But that score needs honest context, in both directions. Webflow sits at 1.4 and Shopify at 1.3 on the same platform, so the low score reflects a category-wide pattern in which satisfied subscribers rarely post and billing disputes reliably do. Trustpilot itself notes the company has not recently invited reviews, so the sample may not be representative. Business-software review sites aimed at buyers rate the product considerably higher than consumer review sites do. The fair reading is that the product's design capability is not what people complain about; its commercial mechanics are. Treat the score as a strong signal about billing and support expectations, not as a verdict on whether the tool builds good websites.
Credit consumption is unpredictable before you commit. No per-action credit consumption rate is published anywhere, and the help article that would explain it is not reachable. You can see how many credits each tier includes, but not how far they go, which makes it impossible to size your plan in advance. Combined with the fact that any one teammate's agent usage draws down the shared allowance, this is the least transparent part of the offering.
Total cost scales faster than the tier price. Running a multilingual site therefore costs you on three axes at once: seats, the add-on and credits. A small team on Pro with several editors and localization can land at several times the advertised $30.
The free tier is an evaluation tool, not a hosting plan. The jump from 1 GB to 50 GB of bandwidth is fiftyfold, which tells you plainly where the free tier is meant to stop.
Code ownership is an unanswered question. The official site makes no statement about code export in any reachable page. Third-party discussion consistently reports that no export exists and treats platform lock-in as a given, and a small market of third-party export tools has grown up around the need — but none of that is an official position. If your organisation requires an exit path, ask the vendor directly before adopting rather than relying on either the critics or the silence.
Some AI capability is not finished. External agents are still labelled a preview and are free only for as long as that preview lasts. A preview feature has neither a settled shape nor a settled price, so do not build a workflow that depends on it.
Service guarantees are tier-dependent. An uptime guarantee is promised only at the Enterprise tier, so businesses on Basic or Pro are running a production site without a contractual availability commitment.
There is a real free tier, but it is designed for evaluation. It includes 500 credits to try, a Framer subdomain and 1 GB of bandwidth — enough to build and judge the tool, not enough to run a trafficked site. A custom domain requires at least the $10 Basic plan.
Four agent surfaces anchor the product: designing on the canvas, running the CMS, writing code, and connecting to outside AI. In practice that means drafting and revising layouts, setting up and organising content collections, and generating code components with working property controls. The output remains editable on the canvas rather than arriving as a fixed artefact.
Credits power the agents and localization, and credits are shared across your entire workspace and its editors. The honest answer to the second half of the question is that you cannot know in advance: no per-action credit consumption rate is published anywhere. Included amounts are 1,000 a month on Basic and 3,000 on Pro, with top-ups available.
More than the tier price. Seats are billed separately at $20 per editor and $10 per content editor, credits can be bought beyond the included allowance, and localization is an add-on rather than an included feature. Budget across all four inputs rather than from the headline figure.
Its Trustpilot score is 1.6 out of 5 across 135 reviews, with 76 percent of them one star, and the complaints concentrate on billing, limits and support rather than design quality. Context matters though: Webflow sits at 1.4 and Shopify at 1.3 on the same platform, so this pattern is common across the category. Read it as a warning about commercial mechanics rather than about whether the builder works.
The official site makes no statement about code export in any reachable page. Third-party sources consistently say no export is offered, but that is not an official confirmation. If an exit path is a requirement for you, get an answer from the vendor in writing before you build on it.
Hosting, security, analytics, CMS and SEO all ship built in rather than assembled from plugins, so the fundamentals are covered without add-ons. Whether the depth of control suits an advanced SEO programme is a separate question, and one worth testing on a small site before migrating an established property that already earns traffic.
Within limits. CMS collections are capped by tier — two on Basic, ten on Pro — with page and item ceilings alongside. Sites with a handful of content types fit comfortably; archives running to tens of thousands of entries fit awkwardly and push you toward Enterprise pricing.
There is no official mobile editing app in the App Store. Editing happens in a desktop browser, which suits a canvas-based design surface. Published sites are of course responsive and display normally on phones.
At the Enterprise tier, yes — that is the direction the company has explicitly pushed. Enterprise brings custom limits, unlimited editors, SCIM and SSO, plus an availability commitment. On lower tiers you are running production without a contractual uptime guarantee, which some organisations will not accept.
