Retool's Business plan costs $50 a month per builder and $15 a month per internal user on annual billing, according to its pricing page read on 28 October 2026. With three builders and 100 users that is $19,260 a year, plus $180 for each extra person. A custom tool has the opposite shape: a large cost up front, then running costs that barely move with headcount.
This page plots both curves from 10 to 500 users, shows the build budget each point can pay for, and covers costs missing from the price lists. Prices come from vendor pages; the curve is our own arithmetic, saved as a raw file you can rework.
How each option is priced
Retool. Retool charges per seat, set by what each person did that month. A builder is "a user who created or edited an app or workflow this billing cycle". An internal user is an employee who uses the apps. External users, such as clients, need Business or Enterprise. Its pricing FAQ says "Retool doesn't store your data": it connects to your databases and APIs.
| Plan | Builder, annual / monthly | Internal user, annual / monthly | Users | What it adds |
|---|---|---|---|---|
| Free | $0 | $0 | Up to 5 | Production environment only |
| Team | $10 / $12 | $5 / $7 | Unlimited | Staging and production |
| Business | $50 / $65 | $15 / $18 | Unlimited plus metered external | Audit logs, rich permissions, external users, embedded apps |
| Enterprise | On request | On request | Unlimited | SSO, SCIM, 2FA, Git source control, granular roles, self-hosting |
Source: Retool pricing page, read 2026-10-28. Vendor-sourced list prices; Enterprise is quoted by sales.
A custom tool. A custom internal tool is an ordinary web application, for us a Next.js front end on Vercel and a Postgres database on Supabase. Vercel Pro is $20 a month with $20 of included usage; its Hobby plan "is for personal, non-commercial use". Supabase Pro starts at $25 a month with 100,000 monthly active users and 8 GB of disk. A tool for 500 staff sits far below both allowances, so hosting stays near $45 a month at any headcount here.
The cost that scales is people: building the tool, then upgrades, patches and the odd new field. The US Bureau of Labor Statistics puts the median software developer wage at $135,980 in May 2025, or about $65 an hour over a 2,080-hour year, before employer costs.
The cost curve by number of users
We held builders at three and changed only the number of users. On the custom side we assumed $45 a month of hosting and eight hours a month of upkeep at $65 an hour. The upkeep figure is our assumption, not a benchmark: it covers work Retool would otherwise absorb, such as upgrades, hosting and authentication.
| Total users | Retool Team, annual | Retool Business, annual | Retool Business, monthly billing | Custom tool, running only |
|---|---|---|---|---|
| 10 | $780 | $3,060 | $3,852 | $6,816 |
| 25 | $1,680 | $5,760 | $7,092 | $6,816 |
| 50 | $3,180 | $10,260 | $12,492 | $6,816 |
| 100 | $6,180 | $19,260 | $23,292 | $6,816 |
| 250 | $15,180 | $46,260 | $55,692 | $6,816 |
| 500 | $30,180 | $91,260 | $109,692 | $6,816 |
Source: Retool, Vercel and Supabase pricing pages and BLS, read 2026-10-28; raw file research/OB-0150-retool-cost-curve-2026-10-28.csv. Excludes the custom build itself, Retool Enterprise, AI credits and usage overages.
The custom column is flat because nothing in it is billed per person. Each Retool column rises by the internal user price: $60 a year per extra user on Team, $180 on Business. The table leaves out the largest custom cost, the build, on purpose.
What the build budget has to fit inside
Read the curve backwards: the yearly gap between Retool and the custom running cost, times the years you will keep the tool, is the most a custom build can cost before it loses money. We used three years and converted the budget into developer weeks at the same $65 an hour.
| Total users | Budget vs Retool Team | Developer weeks | Budget vs Retool Business | Developer weeks |
|---|---|---|---|---|
| 25 | None, Team is cheaper | 0 | None, Business is cheaper | 0 |
| 50 | None, Team is cheaper | 0 | $10,332 | 4.0 |
| 100 | None, Team is cheaper | 0 | $37,332 | 14.3 |
| 250 | $25,092 | 9.6 | $118,332 | 45.3 |
| 500 | $70,092 | 26.8 | $253,332 | 96.9 |
Source: our arithmetic on the cost curve table; raw file research/OB-0150-retool-cost-curve-2026-10-28.csv. Developer weeks are 40 hours at the BLS May 2025 median wage, before employer costs.
Treat the rows as a sense check. Four weeks will not produce a secure tool with logins, roles and an audit trail, so at 50 users Retool Business still wins. At 100 users, 14 weeks covers a focused tool. From 250 users, cost stops being the question.
Team is the harder line to beat: a custom tool needs about 250 users before three years of savings pay for ten weeks of work. But Team lacks audit logs, rich permissions and external users, which is what pushes teams onto Business.
Costs the price lists leave out
Builder creep. Anyone who edits an app that month is billed as a builder. If six people tweak apps instead of three, the Business bill rises by $105 that month.
Plan jumps. SSO, SCIM, two-factor enforcement, Git source control and granular roles are listed as Enterprise features. If policy requires SAML login on every internal app, there is no published Retool price to plan with; a custom tool uses the identity provider you already pay for.
External users. If the tool will face clients, Retool Business includes the first 50 external users free, then charges $8, $6 and $4 a month per user in tiers up to and beyond 500. The page does not say whether the tiers are marginal. For a portal with 1,000 client logins, that reads as roughly $4,000 to $5,100 a month on top of internal seats. On the custom side, the same 1,000 users are 1% of Supabase Pro's included monthly active users.
Self-hosting. Retool lists self-hosting only under Enterprise. Retool's capacity planning guide asks for a platform database with "at least 60GB of storage" and recommends managed Postgres. It solves data residency, not the seat bill.
Lock-in and what you own
A custom tool is code in your repository that any developer who knows the framework can maintain or move.
A Retool app is a configuration of Retool components and queries. Retool's docs say it exports as "a single JSON file or a Toolscript ZIP archive" that "does not contain any resources", so connections are recreated by hand, and it imports only into Retool. Leaving means rebuilding the screens, even though your data never moved.
Open-source builders sit between the two. Appsmith Community is free to self-host; its cloud Business plan is $15 per user a month. Budibase's self-hosted plan is "free to use indefinitely" with unlimited users. Both remove the per-seat slope if you run the server.
Our bias: Octo builds applications in Next.js, so a custom build is work we sell. We have no commercial relationship with Retool, Appsmith or Budibase.
Where speed and AI change the maths
Retool's case is speed: point a table at a database and an operations team has a working screen in an afternoon. That holds for read-and-edit screens over existing data, which describes most admin panels.
The gap is narrower than it was. In Retool's 2026 Build vs. Buy report, a survey of 817 of its own customers and builders, 93% use large language models to code, build or automate, and 51% have shipped production software with AI. The same report found 35% had replaced at least one SaaS tool with a custom build, while 26% named maintenance burden as a blocker. Our internal tools statistics post covers that survey's sample and bias.
AI does not shorten the part that makes an internal tool safe: who sees which rows, who approved a change, what happens when two people edit one record. Retool Business ships audit logs and permissions. A custom tool has to build them, so budget for them from week one.
How to use this
Stay on Retool when the tool has fewer than about 50 users, its builders are not developers, the screens mostly edit existing data, and nobody needs SSO or Git.
Plan a custom tool when you expect 250 or more users, need Enterprise features, want clients to log in, or want the tool in your product's codebase and review process. Between 50 and 250 users, rerun the numbers with your own builder count and developer cost.
Build the exit into either choice. Keep permissions in the database or an API, not only in the screens, so either front end can be replaced screen by screen. Our Supabase vs Firebase comparison covers that backend choice, and our Wized vs custom code piece makes the same argument for logged-in apps on Webflow.
For a custom build we would scope the first version to the two or three workflows that cost the most time, and quote per project. Our web app development service explains how we scope it.
