Open-source Retool alternatives for a Next.js SaaS (2026)

Quick answer (August 2026)

If you are running a Next.js SaaS and want an open-source, self-hostable Retool alternative in 2026, the honest shortlist is five projects, and the right pick depends on one question: do you want a drag-and-drop builder that runs as its own server, or a code-first framework that lives inside your app? For a fast internal admin over your Postgres, Appsmith (Apache-2.0, 40.6k stars) is the most permissively licensed drag-and-drop option, with ToolJet (AGPL-3.0, 38.3k stars) and Budibase (GPLv3, 28.2k stars) as the other two GUI builders. If you would rather own the UI as React code in your existing Next.js repo, Refine (MIT, 35.5k stars) is the code-first pick, and Windmill (AGPLv3, 17.5k stars) is the choice when the "internal tool" is really a pile of scripts and workflows. None of these is a like-for-like Retool clone, and that is the point: you trade some polish for code ownership and a flat hosting bill.

Why teams look past Retool in 2026

Retool is still an excellent product. The reason people end up on a search for alternatives is rarely that Retool is bad; it is the pricing shape and the ownership model.

Retool's published 2026 pricing lists a Team plan at roughly 9 EUR per builder and 5 EUR per internal user per month, and a Business plan at roughly 46 EUR per builder and 14 EUR per internal user per month (retool.com/pricing, 2026). Per-seat pricing is fine at three people. It compounds fast.

Here is the arithmetic nobody puts on their landing page. Take a small ops team on Business: 4 builders and 25 internal users who just need to view and edit records.

• Builders: 4 x 46 EUR = 184 EUR per month
• Internal users: 25 x 14 EUR = 350 EUR per month
• Total: 534 EUR per month, about 6,400 EUR per year

Every new person you add to the internal tool is another line on that bill. Self-hosting one of the open-source builders below runs on a small VPS for a flat cost that does not care whether 5 people or 500 people log in. That single difference, seat-metered SaaS versus flat self-hosted infrastructure, is the real driver behind almost every "open source Retool alternative" search.

A counter-take on the rumor going around: several posts in 2026 claim Retool "removed self-hosting" or made it "Enterprise-only." Retool's own documentation still describes a self-hosted deployment in your own VPC via Docker, so treat the lockout story as unconfirmed. The durable reasons to switch are seat economics and code ownership, not a self-host that vanished.

Two families, not one list

The mistake most "best Retool alternatives" listicles make is ranking all of these on one axis. They are two different kinds of tool.

GUI builders that run as their own server. Appsmith, ToolJet, and Budibase are low-code platforms. You deploy them (usually with Docker), open a visual editor in the browser, drag widgets onto a canvas, and bind them to your database and APIs. For a Next.js SaaS this is a second application you operate next to your app, not code inside it. That is fine, and often preferable, because the internal tool stays cleanly separated from your product.

Code-first frameworks that live in your codebase. Refine is a React framework: you build the admin UI as components inside a repo, and it fits naturally into a Next.js app because it is just React. Windmill sits at the other end, turning scripts in Python, TypeScript, Go, or Bash into internal apps and workflows. You reach for these when you want the internal tool versioned in Git alongside everything else, or when the "tool" is mostly logic.

For the deep three-way on the GUI builders specifically, we compared them in Appsmith vs ToolJet vs Budibase. This piece zooms out to the full Retool-replacement decision.

The five, compared (August 2026)

Stars and licenses are pulled live from each project's GitHub repository on August 8, 2026.

The Next.js and serverless lens

This is the axis the generic roundups skip, and it is the one that matters if you are already on Next.js.

Four of these five run as a persistent server. Appsmith, ToolJet, Budibase, and Windmill all expect a long-running Node.js or Rust process and a database, deployed with Docker. That is a poor fit for a serverless Next.js deployment on Vercel, where there is no persistent process to host them. The clean pattern is to run the builder on its own small box (a VPS, Fly, Railway, or a container host) and point it at the same database your Next.js app uses. Treat it as a sibling service, not something you bolt into your app.

Refine is the exception. Because it is a React framework, it compiles into your Next.js build and deploys wherever your app already deploys, serverless included. If "no extra server to operate" is a hard requirement, Refine is the only one of the five that satisfies it, at the cost of writing the UI yourself instead of dragging it. That trade is the same one we drew in Refine vs React-Admin vs AdminJS.

The tools, honestly

Appsmith is the closest like-for-like Retool experience in open source. Apache-2.0 is the most permissive license here, which matters if you plan to modify it heavily or embed it. It has the largest widget library of the drag-and-drop three and a polished editor. The catch is that it is a heavyweight self-hosted deployment, and very large datasets in the grid can get sluggish. Source: github.com/appsmithorg/appsmith.

ToolJet leans hardest into workflows and AI-agent building, and it markets itself as the AI-native option. It is a strong choice when the internal tool is really an automation with a UI on top. The trade-off is the license: AGPL-3.0 is the strictest copyleft of the group, so read it carefully before you fork or embed. Source: github.com/ToolJet/ToolJet.

Budibase is the fastest path from a database schema to a working CRUD app, and it is the most approachable for non-developers on your team. It is GPLv3 overall, though Budibase notes that the apps you build with it do not package GPL code, so your internal tools are not encumbered. It is less suited to complex, deeply custom applications. Source: github.com/Budibase/budibase.

Refine is not a drag-and-drop builder at all; it is an MIT-licensed React framework for building admin panels and internal tools in code. For a Next.js team it is the most natural fit, because the tool is just part of your app. The cost is real developer time, and there is no visual editor. Note that its public release cadence slowed through 2026 (the repository's last significant push was in June 2026), so weigh momentum alongside fit. Source: github.com/refinedev/refine.

Windmill is the odd one out, and useful precisely for that. It turns scripts into internal apps, workflows, and cron jobs, with a fast Rust core. If your "internal tool" is mostly logic (data pipelines, admin actions, one-off operational scripts) rather than forms over a table, Windmill fits better than any of the GUI builders. Its server code is AGPLv3, with some components under Apache-2.0. Source: windmill.dev/docs/compared_to/retool.

How to choose

• You want the Retool feel, drag-and-drop, minimal code, and the most permissive license: Appsmith.
• The tool is really an automation or AI workflow with a thin UI: ToolJet, or Windmill if it is script-heavy.
• A non-developer needs to spin up CRUD apps over a schema quickly: Budibase.
• You are all-in on Next.js and refuse to operate a second server: Refine, and accept that you write the UI.
• You have five internal users today and will have fifty next year: any self-hosted option here, because the flat infrastructure bill is the whole reason to leave seat-metered SaaS.

There is no single winner, and any roundup that crowns one is selling something. Match the tool to how your team works and to whether you want a separate server or code in your repo.

FAQ

What is the best open-source Retool alternative in 2026?
There is no single best. For a drag-and-drop admin, Appsmith (Apache-2.0) is the most permissively licensed of the GUI builders; ToolJet leans into AI workflows; Budibase is fastest for schema-to-CRUD. For a code-first admin inside a Next.js app, Refine (MIT) is the pick. For script-and-workflow tools, Windmill fits best.

Are these Retool alternatives really free?
The open-source editions are free to self-host under their respective licenses. You still pay for the infrastructure to run them (a small VPS or container host), which is a flat cost that does not scale with the number of users, unlike Retool's per-seat pricing.

Which Retool alternative works best with a serverless Next.js app on Vercel?
Refine, because it is a React framework that compiles into your Next.js build and needs no separate server. Appsmith, ToolJet, Budibase, and Windmill all run as persistent services, so host them on their own box and point them at the same database.

Did Retool make self-hosting Enterprise-only in 2026?
As of August 2026 this is unconfirmed. Retool's own documentation still describes a self-hosted deployment in your VPC via Docker. Treat "Retool killed self-hosting" claims skeptically; the stronger reasons to switch are seat pricing and code ownership.

What is the difference in licenses, and does it matter?
It matters if you fork or embed. Appsmith is Apache-2.0 (most permissive) and Refine is MIT (very permissive). ToolJet and Windmill's server are AGPLv3 (strong copyleft). Budibase is GPLv3 overall, but Budibase states the apps you build do not package GPL code.

Do I have to write code to use these?
Appsmith, ToolJet, and Budibase are low-code drag-and-drop builders, so mostly no. Refine is a React framework, so yes. Windmill is script-first, so you write the logic but it generates the app shell.