Flutter - free resource mobile app, desktop, web

Flutter 3.47 landed on August 13, 2026, and for once the headline is not a feature — it is an architectural split. The design system is moving out of the SDK. Add Impeller becoming the default renderer on every desktop platform and Widget Previews going stable, and this is the most structurally significant Flutter release in years.

The numbers behind it: 1,356 commits from 169 contributors, 66 of them first-timers, shipping as Flutter 3.47.0 • Dart 3.13.0 • DevTools 2.60.0.

Material and Cupertino become packages you choose

The change everyone is talking about: material_ui and cupertino_ui are now standalone packages on pub.dev, both at 1.0. The libraries still live inside the SDK for now, but the SDK copies are scheduled for formal deprecation in the November stable release.

Why this matters more than it sounds:

• Design systems can now ship on their own clock. Instead of waiting for a quarterly SDK release, Material and Cupertino can iterate weekly. Cupertino in particular needs it — see the survey numbers below.
• You can pin your design system independently of your framework version. Upgrading Flutter no longer means swallowing every widget change at the same time.
• Apps that use neither can drop both. If you ship a fully custom design system, the Material tree stops being a mandatory dependency.

Migration is a one-liner:

dart fix --apply --code=migrate_design_widgets

For the messy middle — your app migrated but a dependency still imports the old paths — there is a MaterialUiCompatibilityBridge to keep both worlds talking.

Impeller is now the default on macOS, Windows and Linux

Impeller replaces Skia as the default renderer across all three desktop platforms. The practical payoff is the same one mobile got: shaders are compiled at build time, so shader compilation jank goes away. On macOS, wide gamut color is on by default as well.

There are opt-outs per platform, but the blog is blunt that fallback options will be removed in a future release. Treat the opt-out as a bug-reporting window, not a long-term strategy.

Widget Previews graduate to stable

Widget Preview is no longer experimental. The stable version adds local caching for faster startup and abstract theme APIs, so a preview can be rendered against multiple themes without wiring up a full app shell. If you tried previews in an earlier release and found them slow, this is the version to retry.

The upgrade tax: iOS 15 and macOS 12 minimums

To support Xcode 27, the floors moved:

Apps also need to adopt the UIScene lifecycle for iOS 27 compliance. If you skipped that migration when 3.44 flagged it, this is where it stops being optional.

Web: Wasm keeps inching toward default

The WebAssembly story from the 2026 roadmap continues. You can build with:

flutter build web --release --wasm

New this release: experimental deferred loading for Wasm behind --enable-wasm-deferred-loading. The prerequisite is still the same — migrate off legacy dart:html to package:web before you expect any of this to work cleanly.

On the plugin side, 92 of the top 100 iOS plugins have migrated to Swift Package Manager. The CocoaPods era is ending faster than most upgrade guides assume.

What the Q2 2026 survey says about where Flutter actually hurts

Google published the Q2 2026 survey results the same week — 3,500+ complete responses collected June 8–22, 2026 — and it reads as the subtext for this release.

The good news is unusually good:

• 93% overall satisfaction, with 58% “very satisfied” (up 6 points from Q4 2025)
• 83% trust Flutter to meet their needs, up from 77%
• For the first time, every developer subgroup cleared 90% satisfaction

Then the outlier:

• Cupertino widgets: 61% satisfaction, down 6 points — the lowest-rated area in the survey, against 92% for Dart, 91% for Android, 90% for the core framework.

Which is exactly why decoupling Cupertino into a package that can ship weekly is the fix, not a refactor for its own sake.

The top pain points name the rest of the release, too: platform/ecosystem maturity (44%), driven by version-matrix complexity during upgrades, then tooling and IDE experience (33%), then bugs and stability (24%).

One more number worth sitting with: developers trust community “battle-tested” features at 41% versus 26% for Google-built features, and trust in Flutter runs 20+ points ahead of trust in Google (83% vs 62%).

The AI trend line: agents are now a first-class Flutter workflow

The survey’s AI section is the clearest signal of where the ecosystem is heading. On adoption:

• Claude Code — 32%
• Antigravity — 23%
• GitHub Copilot — 19%
• Cursor — 18%
• Codex — 17%

Agentic tools now lead the pack, ahead of the autocomplete generation. VS Code (66%) and Android Studio (40%) remain the editors those agents run inside.

Google is building for that directly. Skills for Dart and Flutter, announced in May 2026, are task-oriented instruction sets that teach an AI agent how to do a specific Flutter job — the initial five cover integration testing, localization, responsive layouts, pattern-matching refactors, and test coverage. The framing from the announcement is the useful mental model: MCP gives an agent tools; a Skill gives it the blueprint and the professional know-how. They load progressively, only when relevant, which keeps token usage down.

The problem being solved is the one every Flutter developer using an LLM has hit: the framework moves faster than model training data. A release like 3.47, which changes import paths for the two most-used libraries in the ecosystem, is exactly the kind of change a stale model will get confidently wrong.

Your upgrade checklist

1. Run dart fix --apply --code=migrate_design_widgets and move to material_ui / cupertino_ui before the November deprecation.
2. Audit dependencies for packages still importing the SDK design libraries; use MaterialUiCompatibilityBridge while you wait on them.
3. Bump your deployment targets to iOS 15 / macOS 12 and finish the UIScene migration.
4. Smoke-test desktop builds under Impeller — and file bugs now while the Skia fallback still exists.
5. Retry Widget Previews if the experimental version put you off.
6. Test a Wasm build and check whether deferred loading helps your bundle.
7. Point your AI agent at the Dart & Flutter Skills so it stops generating 3.44-era imports.

The bottom line

Flutter 3.47 is the release where Flutter stops being one monolith. Material and Cupertino become packages you version yourself, Impeller finishes its takeover on desktop, and previews become a normal part of the loop. The survey explains the strategy: satisfaction is high almost everywhere, Cupertino is the exception, and shipping design systems on their own schedule is how you fix that without holding the SDK hostage. Budget an afternoon for the migration — but do it before November, when the old imports start warning.

Originally published on FlutterCook. Read the latest version there — that copy is the one kept up to date.

The latest stable Flutter release, Flutter 3.44, is one of the most consequential updates of the year for teams shipping to Apple platforms and the web. The documentation was last refreshed on July 10, 2026, and the release notes read like a checklist of the pain points cross-platform developers have been asking Google to fix.

Full iOS 26 and macOS 26 support

Flutter 3.44 (building on 3.38) fully supports the iOS 26, Xcode 26, and macOS 26 releases. That is the headline for most teams: as Apple shipped its Liquid Glass redesign and a new toolchain, Flutter kept pace so you can build and submit against the current SDK without hacks.

There is one migration to be aware of. Flutter now supports Apple’s mandated UIScene lifecycle, and code migration is required to adopt it. If you maintain an iOS Flutter app, this is the upgrade task to schedule — it is mechanical, but it is not optional going forward.

Cupertino Squircles land

Apple’s rounded corners have never been simple circles — they use continuous curvature, the “squircle.” Flutter 3.44 introduces Cupertino Squircles via a new RSuperellipse shape that matches Apple’s native rounded corners. This required deep changes in the rendering engine, and it closes one of the last visible gaps between a Flutter iOS app and a native one. On iOS 26, where Liquid Glass makes shape and light more prominent, getting corners exactly right matters.

Stateful hot reload becomes the default on web

For anyone who builds Flutter Web, this is the quality-of-life win of the release: stateful hot reload on web is now the default. You no longer lose your navigation stack, form inputs, or scroll position when you hot reload. It brings web development to parity with the native mobile experience Flutter developers have enjoyed for years — a genuinely different day-to-day feel.

Security: Android screen-share protection

On Android, developers can now protect sensitive content when users share their screen — hiding customer information, account numbers, and other private data from screen recordings and casts. For fintech, healthcare, and banking apps, this is a feature that previously required native platform code and now ships in the framework.

What to do with this release

• Schedule the UIScene migration if you ship to iOS. It is required for future compatibility.
• Adopt RSuperellipse for cards, buttons, and sheets on iOS to match native corner geometry.
• Re-test your web build with the new hot reload default — the workflow improvement is significant enough to change how you develop.
• Add screen-share protection to any screen that shows sensitive data on Android.

The bottom line

Flutter 3.44 is a “catch up to the platforms and polish the edges” release, and that is exactly what mature cross-platform frameworks should ship. Full iOS 26 and macOS 26 support keeps you current with Apple; Squircles and screen protection close native gaps; and stateful web hot reload quietly makes Flutter Web a much nicer place to work. If you have been putting off an upgrade, 3.44 is worth the afternoon.

Originally published on FlutterCook. Read the latest version there — that copy is the one kept up to date.