The visual editor library ecosystem in 2026: what our dataset tells us

What this report is

This is an analysis of our own dataset rather than a survey of opinions. Every figure below is computed from the 34 product records published at /api/libraries.json, and you can recompute any of them yourself.

The point is not to rank products. It is to describe the shape of the category, because the shape explains most of the confusion buyers experience.

The category is four categories

Products marketed with the same words do four different jobs:

Embeddable libraries (13). You install them from npm and mount them in your application. Your data, your hosting, your interface work. GrapesJS, Puck, Craft.js, Editor.js, Plate, Tiptap, Lexical, React Page, Silex, CKEditor 5, TinyMCE, BlockNote and Quill.

Commercial SDKs (9). Licence-key products supplying a finished editor plus infrastructure. Builder.io, Plasmic, TeleportHQ, Unlayer, Beefree SDK, Stripo, Topol, the GrapesJS Studio SDK and PageKit occupy this space with very different shapes.

Headless CMSs with visual editing (4). Storyblok, Sanity, Contentful Studio and Directus. Their editors serve your content team, not your customers.

Hosted platforms (8). Webflow, Framer, Elementor, Wix Studio, Duda, Brizy, Sitejet and Zillapage. Mostly not embeddable at any price; present because every build-versus-buy conversation compares against them.

The single most common evaluation error we see is comparing across these groups on features. A CMS and an embeddable library can both be described as "visual editing" and neither can do the other's job.

The two constraints that decide most shortlists

Self-hosting. 18 of 34 can run entirely on infrastructure the buyer controls. If page content cannot leave your network, nearly half the category is gone before features are discussed.

Resale under your own brand. 20 of 34 can be shipped to your own customers as your product. The best-known names — Webflow, Framer, Builder.io, Plasmic — are not among them, which is why the search for a "white-label website builder" is so frustrating and why the white-label category page exists.

Answer these two questions first and the shortlist usually resolves to three or four candidates.

Pricing has stopped being about price

The email sub-category illustrates this most sharply. Four products with materially similar capabilities charge:

• around $60 per month, metered by customer accounts that open the editor;
• $100 per month, metered by unique templates created;
• $250 per month, a flat platform tier;
• $350 per month, a platform tier plus usage components.

A product with 2,000 light accounts and a product with 40 heavy accounts will find these four in completely different orders. The capability differences between them are small; the metering differences are decisive.

This generalises beyond email. Builder.io moved to credit-based pricing that meters work done rather than seats occupied. Storyblok meters spaces, which punishes agencies. Directus meters your company's revenue. Framer and Webflow charge for seats separately from plans.

The practical conclusion: model the meter against your own shape before comparing headline prices. Our build-versus-buy calculator does the licence-versus-hours half of this; the meter half needs your own numbers.

What every product in the dataset refuses to do

Three capability gaps recur across the whole category, and they are worth knowing because they will not be filled by choosing differently.

Nothing server-renders. Every editor here mounts against a live DOM. Editing is always a client-side, authenticated concern, and any architecture that assumed otherwise needs revising.

Nothing gives you both React-native components and portable HTML output. Component-tree editors store data that means something only inside your app; HTML editors cannot render your React components in the canvas. Teams that need both — a page builder and an email builder, say — run two editors, and that is the correct answer rather than a compromise.

Nothing includes the surrounding product. Asset management, permissions, versioning and publishing are yours in every open-source option and vary widely in the commercial ones. Our model puts that work at roughly 480 hours, and it is the single most underestimated line in editor projects.

Measurement findings

We measured what we could rather than what would make a better chart.

Bundle sizes span 13.5×, from Craft.js at 29.2 kB gzip to TinyMCE at 393.9 kB, and the spread tracks how much editor interface each library includes rather than how capable it is. See the bundle benchmark.

Time to a rendered editor spans 14.3×, from 70 ms to 1004 ms, and again the ordering tracks how much appears on screen. See the time to first editor benchmark.

Only 12 of 34 products could be measured at all. The rest are hosted services behind licence keys or platforms with no local build. That is the clearest structural fact about this category: most of it cannot be independently benchmarked by anyone, which is precisely why vendor claims dominate the conversation.

What we cannot tell you

This report analyses products, not usage. We do not know which of these is most deployed, how satisfied their users are, or how often teams abandon an editor project — because nobody has published that data and we have not collected it.

That is why the State of Visual Builders survey exists as an open survey rather than as a page of numbers. When it has enough responses to be worth reporting, the results will appear there with the sample size attached.

How to cite this

EditorStack (2026). The visual editor library ecosystem in 2026. Computed from the EditorStack dataset, 19 August 2026. https://www.editorstack.cc/research/editor-library-ecosystem-report

The dataset is CC BY 4.0. See how to cite for BibTeX.