← Default view
Itenium.Forge · itenium-ui procurement record File IF/2026/UI-01 Opened 14 Jul 2026 Depth survey · 31 exhibits

Four subjects.
Three categories.
One cash gate.

Maturity and governance baseline on the four React UI kits under evaluation for a greenfield admin console — Nx 21, React 19, Vite, Tailwind 4. Stars, versions, licence, funding, bus factor and issue throughput, each entered as a record.

Subjects
Material UI · shadcn/ui · Mantine · Base UI
Data cut
14 Jul 2026 — GitHub REST API, npm registry, vendor filings
Classification
Open source · unrestricted · one paid tier (see Exhibit D)
Filed under
Which React UI component kit should itenium-ui adopt in 2026?
Abstract cover plate: stacked slabs with one highlighted band
Plate 1 — case cover, IF/2026/UI-01
Finding — baseline layer only 14.07.2026

These four are not four options in one category — they are three categories. Material UI is a batteries-included library with a paid data grid: the row-grouping / aggregation / Excel-export tier an admin console actually wants costs $599/dev/yr [15]. Mantine is the same shape but 100% MIT — with a 94% single-maintainer bus factor [3] and an $18.5k/yr budget [25]. shadcn/ui is a code-distribution registry, not a dependency, and as of July 2026 it initialises new projects on Base UI, not Radix [21]. Base UI is the primitive layer under that — MUI-staffed, built by the ex-Radix / Floating-UI team, stable since Dec 2025 [20].

Picking "shadcn/ui" in 2026 therefore means picking MUI's primitives with none of MUI's licensing.

Category collapse confirmed
Exhibit A

The maintenance inversion

GitHub REST API · 14 Jul 2026

Popularity and health point in opposite directions. The most-starred React UI repo on earth is also the worst at draining its own inbox; a project with a quarter of its stars and one maintainer runs an effectively empty tracker. Read left of the name for reach, right of it for debt.

← ⭐ Stars Subject Open issues → 90-day throughput
⭐ 119k
shadcn/uicode registry
1,019
0.35× ✗
⭐ 99k
Material UIlibrary
1,387
1.82× ✓
⭐ 31k
Mantinelibrary
12
1.04× ✓
⭐ 10k
Base UIprimitives
311
0.74× ⚠
⭐ 5.8k
MUI Xpaid grid
1,552
paid support

shadcn/ui closes ~1 issue for every 3 opened (53 closed / 150 opened, 90d) [2]. Mantine — one person — runs a 12-issue tracker and closes more than arrives [3]. Material UI is the only one shrinking a large backlog at speed (206 closed / 113 opened) [1]; Base UI's is growing slowly, which is what a young post-1.0 project looks like [4]. Star count is a popularity signal, not a maintenance signal. Do not let it drive this decision.

Exhibit B

Subject files

Four records, one page each
Subject 01 · MUI SAS, Paris Risk ⚠ commercial

Material UI ⭐ 99k

Batteries-included, styled (Material Design) · runtime dependency
Version
9.2.0 (3 Jul 2026) — v8 does not exist, v7 → v9 to re-align with MUI X [17]
Stable since
v9 on 8 Apr 2026 · monthly minors · v5/v6/v7 tags alive
React peer
^17 || ^18 || ^19 [7]
Styling
Emotion — still a hard peer dep; Pigment CSS paused in alpha [18]
Licence
Core MIT, "free forever" · MUI X Pro/Premium paid [16]
Backer
MUI — company, Paris, founded 2014, 40+ remote staff in 20+ countries [26]
Funding
Bootstrapped — no VC raised. Revenue = MUI X licences [27]
Bus factor
oliviertassinari 22% of top-100 [1]
Tracker
1,387 open · 113 in / 206 out (90d) · shrinking
Downloads
38.7M/mo (Jun 2026), from 27.5M in Jan — largest in field [11]
Governance risk The free tier is the funnel for the paid grid. Products get shelved when the funnel doesn't need them: Joy UI on hold, Toolpad unmaintained, MUI Sync dead [18].
Subject 02 · Vercel-salaried, MIT Risk ✗ throughput

shadcn/ui ⭐ 119k

Code registry + CLI — "this is not a component library" [22]
Version
CLI 4.13.0 (3 Jul 2026) — a devDependency, not a runtime [10]
Delivery
shadcn add copies source into your repo — you own the code
Runtime
Base UI since Jul 2026. Radix survives behind init -b radix, not deprecated [21]
Styling
Tailwind — no cascade negotiation, no theme object
Licence
MIT, nothing gated [2]
Backer
Vercel — shadcn is a design engineer there since Aug 2023; project predates the hire [28]
Funding
A salary. No monetisation of the registry itself.
Bus factor
shadcn 73% of top-100 — but on the registry, not on your files
Tracker
1,019 open · 150 in / 53 out (90d) · growing ~3×
Downloads
22.1M/mo CLI, 6.6× since Jan [14]
Governance risk If shadcn stops, the components are already source files in your repo. You lose future registry updates and the CLI — not your app. The runtime risk transfers to Base UI, which is company-staffed.
Subject 03 · Vitaly Rtishchev, sole Risk ✗✗ bus factor

Mantine ⭐ 31k

Batteries-included, styled (own design language) · runtime dependency
Version
9.4.1 (28 Jun 2026) · 9.0 on 31 Mar 2026 · ~2-week minors [23]
React peer
^19.2.0 only — no React 18 escape hatch [9]
Styling
CSS Modules + CSS variables — no runtime CSS-in-JS
Licence
MIT, "free forever for everyone" · no paid tier at all [24]
Backer
One person. No company, no foundation [24]
Funding
Open Collective — ~$18.5k/yr est. budget, $18.4k raised total, 125 contributors [25]
Bus factor
rtivital 94% of top-100 (13,701 commits) [3]
Tracker
12 open · 80 in / 83 out (90d) · zero-inbox
Surface
120+ components, 70 hooks, 12k Discord · ships @mantine/mcp-server
Downloads
8.0M/mo, 1.8× since Jan — below baseline inflation [12]
Governance risk The maintenance winner and the bus-factor loser, simultaneously. If the maintainer stops, @mantine/core goes stale inside node_modules and you have a fork-or-migrate problem across every screen — nothing has been copied into your repo.
Subject 04 · MUI-staffed, ex-Radix Risk ⚠ young

Base UI ⭐ 10k

Unstyled headless primitives · runtime dependency, no styles to own
Version
1.6.0 (18 Jun 2026) — monthly minors, no majors since 1.0 [20]
Stable since
1.0.0 on 11 Dec 2025 — 35 components at 1.0 [19]
React peer
^17 || ^18 || ^19 [8]
Styling
None. Bring your own.
Licence
MIT, nothing gated
Backer
MUI — dedicated engineers, designers and managers; explicitly positioned against post-acquisition uncertainty at Radix [19]
Authors
ex-Radix + Floating UI + Material UI
Bus factor
atomiks 21% · michaldudak 11% — the flattest in the field [4]
Tracker
311 open · 159 in / 117 out (90d) · growing slowly
Downloads
24.9M/mo, 14× since Jan — the shadcn default switch [13]
Shared exposure Base UI is load-bearing for two of the four options — a standalone candidate and shadcn/ui's default runtime — and MUI X already depends on @base-ui/utils internally. Its health is a shared dependency, not an isolated one.
Exhibit C

The category collapse

Why four candidates are three things

The July 2026 realignment is the single biggest fact of the year for this file: npx shadcn init now selects Base UI, docs show Base UI examples first, and Radix stays supported but demoted [21]. The "vendor-neutral, no-lock-in" path and the "corporate MUI" path now share a runtime maintainer.

▣

Vendored library

Material UI · Mantine

A runtime dependency ships the components. You import them; you do not own them. Upgrades arrive through package.json. Styling comes with the box, and must be negotiated with.

⎘

Code registry

shadcn/ui

A CLI copies source files into your repo. Not a dependency — a delivery mechanism. You own every component after the first add, and future updates are opt-in, not automatic.

⊹

Primitives

Base UI

Behaviour, accessibility, focus management — no styles. The substrate the registry now sits on by default, and the layer both other categories increasingly route through.

"shadcn/ui vs Base UI" is not a real fork in the road — shadcn/ui is Base UI plus Tailwind styling plus a codegen CLI. The real fork is owned code (shadcn + Base UI) vs vendored library (MUI / Mantine).
Exhibit D

The cash gate

MUI X pricing · the only hard commercial gate in the field

Three of four are pure MIT with nothing gated. Material UI's core is MIT too. But the grid features an admin console actually needs sit behind MUI X — and an admin console walks straight into them.

Licence is perpetual + 12 months of maintenance; renewals may rise up to 7%/yr [15]. Seats = concurrent developers touching front-end code that uses Pro/Premium. 30-day non-production trial without a key [16].
TierPer dev / yrWhat it unlocks
Community (MIT) $0 Basic Data Grid, Date Pickers, Charts, Tree View, Scheduler
Pro $299 Multi-filtering, multi-sorting, column resizing, column pinning, date/time range pickers, advanced charts, tree-view drag-and-drop
Premium ← the gate $599 Row grouping · aggregation · Excel export — i.e. table stakes for an admin console
Enterprise $1,399 Premium + guaranteed response times, CSM · min. 15 seats
Enforcement — runtime, not build-time

A missing env var ships a watermark

Validation happens at render via LicenseInfo.setLicenseKey(). On failure the component renders a watermark and logs a console warning in production, not just in dev. There is no build-time check to catch it in CI [16].

Translation for itenium-ui

$599 × every front-end dev, forever

The moment the console needs grouped rows or an Excel-export button, MUI becomes a recurring line item. It is the single largest cost delta between the candidates — and the one none of the other three can charge you, because none of the other three ships an enterprise grid at all.

Exhibit E

Version register & known traps

Latest as of 14 Jul 2026
PackageLatestStable sinceCadenceReact peerStyling engine
Material UI 9.2.0 · 3 Jul 26 [7] v9 · 8 Apr 26 [17] Monthly minors ^17 || ^18 || ^19 Emotion (peer)
MUI X 9.9.0 · 9 Jul 26 [6] v9 · Apr 26 Weekly–biweekly ^17 || ^18 || ^19 Emotion via MUI
Mantine 9.4.1 · 28 Jun 26 [9] 9.0 · 31 Mar 26 [23] ~2-week minors ^19.2.0 only CSS Modules + vars
Base UI 1.6.0 · 18 Jun 26 [8] 1.0.0 · 11 Dec 25 [20] Monthly minors ^17 || ^18 || ^19 None — bring your own
shadcn CLI 4.13.0 · 3 Jul 26 [10] n/a Weekly-ish Follows your project Tailwind
Trap 01 — the missing major

Material UI v8 does not exist

It went v7 → v9 to re-align its major with MUI X v9, which was already there [17]. Expect stale blog posts and LLM output to confidently reference a v8 that was never published.

Trap 02 — the frozen scope

The dead @base-ui-components package

Base UI renamed at 1.0: @base-ui-components/react → @base-ui/react [19]. The old scope is frozen at 1.0.0-rc.0 and still pulls ~400k downloads/week [31] — a live trap for anyone copying pre-2026 snippets.

Trap 03 — the roadmap promise

MUI's Tailwind story isn't shipped

Emotion is still a hard peer dep in v9 [7]. MUI's own roadmap files Emotion-independence and "better integration paths for teams using Tailwind CSS" as future work, and Pigment CSS is paused in alpha [18]. A commitment, not a feature.

Exhibit F

npm downloads — evidence, contaminated

Jan 2026 → Jun 2026, monthly
⚠ Read as ratios, not as growth — baseline is inflated
PackageJan 2026Jun 2026MultipleReading
@mui/material 27.5M 38.7M 1.4× Still the largest single library, by a wide margin [11]
@base-ui/react 1.75M 24.9M 14× ~4× the inflated baseline — the shadcn default switch is doing this [13]
shadcn (CLI) 3.3M 22.1M 6.6× Real, but a CLI install ≠ a runtime dep [14]
@mantine/core 4.5M 8.0M 1.8× Below baseline — holding, not gaining [12]
@radix-ui/react-dialog
the control		 65.8M 247M 3.8× Grew 3.8× while being displaced as shadcn's default. That is not adoption — it is CI, mirrors and AI agents re-installing [30]
Survey data agrees on the shape: State of React 2025 (~2,600 respondents) puts MUI still #1 in usage with shadcn/ui growing fast and close to overtaking it — and notes a large share of respondents use no component library at all. The space is not settled [29].
Referred on

What this baseline hands to the other angles

  1. Base UI is load-bearing for two of the four options — standalone candidate and shadcn/ui's default runtime. Its health (⭐ 10k, 311 open issues, MUI-funded, 7 months post-1.0) is a shared dependency, not an isolated one.
  2. MUI's Tailwind story is a roadmap item, not a shipped feature, and Emotion is still a hard peer dep in v9 — referred to the Tailwind-4 interop angle.
  3. Mantine is the maintenance winner and the bus-factor loser simultaneously. Both facts are real; the risk sits entirely on one person's continued interest.
  4. The only recurring cash cost in the field is MUI X Premium at $599/dev/yr, triggered by exactly the grid features an admin console needs — referred to the component-coverage angle.
Cross-refs

Related files

Parent case
Which React UI component kit should itenium-ui adopt in 2026? Open →
survey · this file

Candidate profiles, maturity & governance

Stars, versions, licence, funding, bus factor, issue throughput.

expedition

Styling engine & Tailwind 4 interop

Tailwind 4 puts every utility in a cascade layer — any kit shipping unlayered CSS silently outranks it.
survey

Admin-console component coverage

Only Mantine covers a CRUD console in-box with a single gap — the data grid.
survey

A11y & primitive foundations

The real choice is which primitive foundation you inherit — and none of the four publishes a VPAT.
survey

Theming, tokens & brand reach

The token substrate decides the kit: @theme vs a JS theme object.
recon

Nx 21 build ergonomics

The copy-in model works in a shared Nx library — if you own the plumbing.
Register

Exhibits & sources

31 entries
01mui/material-ui — repo health, 98,582 stars, 1,387 open issuesofficial · ⭐ 99k 02shadcn-ui/ui — 119,064 stars, 1,019 open issues, 150 in / 53 outofficial · ⭐ 119k 03mantinedev/mantine — 12 open issues, rtivital 94% of top-100official · ⭐ 31k 04mui/base-ui — 311 open issues, atomiks 21% / michaldudak 11%official · ⭐ 10k 05radix-ui/primitives — maintained by WorkOS post-acquisitionofficial · ⭐ 19k 06mui/mui-x — 1,552 open issues; data grid 9.9.0 (9 Jul 2026)official · ⭐ 5.8k 07npm @mui/material — 9.2.0, Emotion peer deps still requiredofficial 08npm @base-ui/react — 1.6.0, no styling dependencyofficial 09npm @mantine/core — 9.4.1, hard react ^19.2.0 peerofficial 10npm shadcn — CLI 4.13.0, a devDependency not a runtimeofficial 11npm downloads — @mui/material 38.66M, Jun 2026official 12npm downloads — @mantine/core 8.04M, Jun 2026official 13npm downloads — @base-ui/react 24.90M, Jun 2026 (14×)official 14npm downloads — shadcn CLI 22.11M, Jun 2026official 15MUI pricing — Pro $299, Premium $599, Enterprise $1,399/dev/yrofficial 16MUI X licensing — runtime enforcement, watermark in productionofficial 17Introducing MUI v9 — "no Material UI v8, like there is no v2"vendor blog 18MUI 2026 and beyond — Pigment CSS paused, Joy UI on holdvendor blog 19InfoQ — Base UI 1.0: 35 components, MUI-staffed, package renamednews 20Base UI releases — 1.0.0 on 11 Dec 2025, monthly minors sinceofficial 21shadcn/ui changelog — Base UI is the default; Radix not deprecatedofficial 22shadcn/ui docs — "This is not a component library"official 23Mantine 9.0.0 — React 19.2+ required for all @mantine/* packagesofficial 24Mantine — 120+ components, 70 hooks, built by Vitaly Rtishchevofficial 25Open Collective — Mantine's ~$18,540/yr estimated annual budgetofficial 26MUI about — founded 2014, 40+ remote staff across 20+ countriesofficial 27Crunchbase — MUI has raised no venture funding; bootstrappednews 28shadcn — joining Vercel as a design engineer (Aug 2023)forum 29State of React 2025 — MUI #1 in usage, shadcn/ui close behindnews 30npm downloads — @radix-ui/react-dialog 247M: the inflation controlofficial 31npm @base-ui-components/react — frozen scope, still ~400k/wkofficial
File IF/2026/UI-01 · candidate profiles, maturity & governance Depth survey · 31 exhibits · ~11 min Canonical record · Parent case · Atlas