KDE Plasma 6.7 Is Coming June 16th — And I Cannot Wait

· 8 min read

KDE shipped the Plasma 6.7 Beta on May 14, 2026. The final release lands on June 16. I’m still on Plasma 6.6.5 — the stable that shipped in Fedora 44 — and I’m going to stay there until 6.7 hits Fedora’s repos. But I have been reading every announcement, every “This Week in Plasma” post, and every release-note dump KDE has been putting out, and I want to say this clearly:

This is the most genuinely exciting Plasma release I can remember in years.

Not because of one headline feature, but because almost every single change on the 6.7 list lines up with something that has bothered me for a long time — and a few of them target problems I had given up on. Let me walk through it.

Union — the theming engine I’ve waited a decade for

The single biggest item in Plasma 6.7 is also the most architectural. Union is a brand-new style engine arriving as a first public tech preview. It is, finally, an honest answer to the long-running fragmentation problem in the KDE ecosystem: QtWidgets apps are styled one way, QtQuick / Kirigami apps another, Plasma shell components a third, and visual consistency across the desktop has depended on multiple separate theming subsystems pretending to agree with each other.

Union takes CSS — yes, real CSS, the same language that styles the web — and processes it into consistent styling for both QtWidget and QtQuick applications. One source of truth, one set of rules, one consistent look across every KDE app, every Plasma widget, every dialog. CSS files will live in /usr/share/union/css/styles/breeze with per-user overrides at ~/.local/share/union/css/styles/breeze.

When you install the union package in 6.7 and re-launch System Settings, Plasma System Monitor, Discover, Spectacle, NeoChat, Haruna or Plasma widget configuration dialogs, they all start rendering through Union. The deliberate goal for this first preview is parity, not reinvention — the dev team wants the Union version to look as close to the existing styling stack as possible, so any visual discrepancies stand out as bugs to file rather than design changes to debate.

That’s exactly the right strategy. You don’t rewrite the foundation of a theming system and redesign the look at the same time. You ship parity, build trust, and then iterate. This release is the foundation-laying step. The next several Plasma releases get to build on top of it.

If you’re brave enough to be on the beta, you can launch any QML/Kirigami app with the legacy style for comparison:

QT_QUICK_CONTROLS_STYLE=org.kde.desktop systemsettings
QT_QUICK_CONTROLS_STYLE=org.kde.union   systemsettings

Run those back-to-back and you’re staring at the next decade of KDE styling.

KWin Vulkan — and what it could mean for NVIDIA users

KWin has been an OpenGL compositor for its entire life. In Plasma 6.7, the Vulkan rendering backend continues maturing, with the DRM backend gaining direct GPU-to-CPU texture transfer, image-format conversion drivers, and extensive testing on modern NVIDIA hardware including the RTX 5070 Ti with the proprietary driver.

I just wrote a field report on running NVIDIA on Fedora KDE Wayland, so I’m going to be specific about why this matters: with an NVIDIA GPU as the primary card, early Vulkan backend testing showed actual performance improvements over OpenGL. With AMD or Intel as primary, performance was roughly equivalent — but the NVIDIA case is exactly where KWin has historically been weakest. A Vulkan renderer that plays better with the NVIDIA driver’s modern paths is the thing my hardware combination needs.

There’s more in the same area: the System Monitor will correctly read NVIDIA GPU utilization again after a regression, and the Wayland stack as a whole gains better protocol error handling during resolution changes — a class of bug that historically affected NVIDIA setups disproportionately.

The combination of explicit sync (landed in earlier releases), the Vulkan backend (maturing in 6.7), and the open kernel modules being mandatory for Blackwell (where the driver work goes) is finally pulling NVIDIA-on-Wayland from “acceptable” toward “actually preferred.” Sitting on 6.6 and watching this land is genuinely exciting.

Plasma Bigscreen is back

Plasma Bigscreen — KDE’s 10-foot UI for TVs, HTPCs, and set-top boxes — returns as a first-class module in 6.7. The pitch is exactly what it sounds like: Plasma technologies, rebuilt for a TV display controlled by a remote or a controller from across the room.

The use cases the community is hinting at: docked Steam Deck connected to a TV, a future Steam Machine running KDE, or any HTPC build where you want a Linux desktop that doesn’t look ridiculous on a 65-inch screen viewed from a couch. I don’t have an HTPC build at home right now, but the moment Plasma Bigscreen is solid I have a Raspberry Pi and an old TV that suddenly look very interesting.

The fact that KDE is willing to bring back and modernize a module like this — rather than declaring “the desktop is the desktop” — is the ecosystem energy I love to see.

Fractional scaling that actually agrees with the pixels

This one is buried in the release notes but it’s huge if you run a HiDPI display. Plasma 6.7 implements the experimental xx-fractional-scale-v2 Wayland protocol, which establishes accurate links between logical coordinates and actual pixels on screen.

In plain English: those small gaps you sometimes see between maximized windows and panels at fractional scales like 125% or 150% — the half-pixel rounding errors that nobody can quite pin down — go away. Maximized windows snap exactly to panel edges. Window borders don’t blur. Click targets land where the cursor says they land.

The fact that fractional scaling is “supported” on Wayland has been true for a while. The fact that it’ll be mathematically correct to the pixel is new in 6.7, and it’s the difference between fractional-scaling-as-feature and fractional-scaling-as-default.

Quality of life — the small changes that pile up

The release notes for 6.7 are full of small things that together change how the desktop feels day-to-day:

  • A light/dark mode toggle in the Brightness & Color widget — one click on the panel to flip the whole desktop. I switch between modes constantly depending on the time of day and how much sun is hitting the room. This is the right place for the switch.
  • A global push-to-talk shortcut — hold a key, microphone unmutes, release, it mutes again. I have spent years gluing this together with scripts and pactl invocations. Built-in is better.
  • Notifications now slide in from off-screen instead of fading in. Subtle, but the motion catches your eye in the periphery without the jarring opacity flash the old animation had.
  • The virtual keyboard’s appearance triggers are configurable — multiple selectable triggers in System Settings, instead of a single hard-coded heuristic. Important for tablet-mode and touchscreen users.
  • “Keep Above Others” moved to the top level of the titlebar context menu instead of being buried in a submenu. I use this constantly to pin a terminal or a reference window over an editor; the round-trip through a submenu was always annoying.
  • Gwenview replaces GIMP as the default SVG viewer — sane. GIMP loading just to look at an icon was always overkill.
  • Kickoff handles rapid-typing-then-Enter correctly, eliminating the small race where pressing Enter before the search results finished populating launched the wrong thing.

None of these are headline features. Together, they’re the kind of polish that turns “the desktop works” into “the desktop feels right.”

Privacy and capture controls, done thoughtfully

Two changes here that I’m particularly happy to see coming.

The old “Hide from Screencast” feature — which let you mark a window so it wouldn’t appear in screen recordings — has been expanded and renamed to “Hide from Screenshots and Screen Recordings”. Now it covers both static screenshots and dynamic captures. If you’ve ever taken a screenshot for a blog post or a bug report and had to manually crop out your password manager or a sensitive terminal, this is the feature for you.

Separately, applications that need to control input devices (think: remote desktop, accessibility tools, certain games) can now be granted persistent authorization instead of re-authorizing every single time the app starts. The right model: one trust decision per app, remembered.

Smart card unlock workflows also got several improvements. Not applicable to me personally, but the fact that KDE keeps investing in enterprise-shaped auth flows on the Linux desktop is meaningful.

What I’ll be watching for at the final release

The final 6.7 release on June 16 will refine the Union baseline based on bug reports from the beta period, push KWin’s Vulkan work a little further, and ship Plasma Bigscreen in a state where I’ll actually try it on a TV. The two betas (May 14 and a second one on May 28) are where the most aggressive polishing happens.

When 6.7 hits Fedora’s repos, I have a checklist for the day I move my main workstation over:

  1. Union toggles cleanly on System Settings, Discover, Spectacle without a single visual discrepancy I’d file as a bug.
  2. KWin Vulkan backend works end-to-end on my RTX 5070, including external HDMI output and external monitor hotplug.
  3. Fractional scaling at 125% on an external display has zero visible gap between maximized windows and the panel.

If the second beta and the early stable reports check those three off, I’m on 6.7 the day Fedora ships it.

Why I’m excited

The Linux desktop spent fifteen years being the thing you put up with to avoid the alternatives. Plasma 6.6 already turned that corner for me — it’s why I’m on Fedora KDE today. Plasma 6.7, on paper, is the release that takes the corner and keeps accelerating: Union is the right architectural bet, KWin on Vulkan is the right path, Plasma Bigscreen returning is the right cultural signal, and the pile of quality-of-life changes is the right tax to be paying down.

This isn’t a fireworks release. It’s a foundation release — the kind that doesn’t dominate headlines but quietly changes the trajectory of the next two or three releases after it. Those are the ones I get most excited about, because they’re the ones you’re still thanking the developers for three years later.

June 16, 2026. I’m counting the days.