WordPress 7.0: The Biggest Update Since Gutenberg

WordPress 7.0 dropped on May 20, 2026. And no, this is not one of those point releases where you skim the changelog, shrug, and hit update. This one actually changes things.

If you were around for the Gutenberg switch in 2018, you know what a “things look different now” release feels like. 7.0 is that. Not because of one single feature, but because of what it adds up to — a native AI layer, a redesigned admin, block-level collaboration tools, and a general sense that WordPress finally caught up to where web development actually is.


The context matters here

2025 was a rough year for WordPress. Legal drama, contributor walkouts, compressed release schedule, loud arguments about who controls what. If you followed any of it, you probably had some uncertainty about where the project was heading.

So 7.0 shipping as cleanly as it did, with features that were actually promised actually landing, feels like a reset more than a release. Worth acknowledging before we get into the features.


1. The WP AI Client

This was the feature I was watching most closely, and it’s the right call.

WordPress 7.0 ships a standardized AI layer — the WP AI Client and Abilities API. The idea: configure your AI provider once (OpenAI, Anthropic Claude, Google Gemini, whatever), and everything that needs AI just uses it. No more every plugin rolling its own integration with its own settings page and its own API key field.

There’s a new screen at Settings > AI Experiments where you opt into specific features: excerpt generation, alt text for images, image generation, content summarization. None of it is on by default. You turn on what you want.

For developers, the Abilities API is the part that opens things up. You can build plugins that hook into the AI layer without duplicating any of the provider plumbing. That’s actually useful infrastructure, not a headline feature bolted onto a press release.

I’ve been critical of how a lot of platforms handle AI integration — mostly dumping it into the UI without much thought. This approach is saner than most.


2. DataViews

The admin list tables have been a sore spot for years. Sort by date? Full page reload. Filter by category? Another reload. Bulk edit and navigate back? Gone, start over. It worked. Nobody was excited about it.

DataViews is the React-based replacement. Filtering, sorting, and bulk editing happen inline with no reload. You can switch between table, grid, and list views on the same screen. There’s also an Activity layout that renders content in a timeline style, which is actually how I’d want to review a content calendar.

The custom views feature is where it clicks for heavier users. You save a view configured exactly how you want — say, a grid filtered to image-heavy posts, or a list scoped to one author’s drafts — and it persists. No re-configuring every time you open the screen.

The old list tables are still there for compatibility. But if you spend real time in the WP admin, DataViews will change how that feels pretty quickly.


3. New blocks (and overdue fixes)

Two new core blocks ship in 7.0.

The breadcrumbs block has been requested since Full Site Editing launched. It’s here now — drop it anywhere in your template, no plugin needed, respects your site hierarchy. Simple, but something I’ve had to solve with a plugin on basically every FSE project.

The icons block adds inline SVG support natively. Built-in library, size and color inline. Again, something that required a plugin before.

The existing block improvements are less flashy but more useful day-to-day:

The Cover block now takes an embedded video as background, not just a video URL. The Grid block is properly responsive now (which it should have been at launch, honestly). The Gallery block supports lightbox natively — click an image, it opens fullscreen, no plugin required.

None of these are the kind of things that make a launch post headline. But they’re the kind of fixes that remove friction you stopped noticing because you’d worked around it for so long.


4. Block-level Notes

Real-time multi-user collaboration was supposed to be in 7.0. It got pulled on May 8 — two weeks before launch — because the HTTP polling sync provider wasn’t stable enough for a core release.

What did ship is block-level Notes. You can leave comments on individual blocks inside the editor, directly, like Google Docs comments but in WordPress. For any editorial workflow where more than one person is touching content before it goes live, this is actually more immediately useful than real-time sync would have been.

The real-time collaboration feature is still coming. WordPress 7.0 exposes the hook for it — hosts and plugins can add WebSocket support when it’s ready. But for now, Notes is what shipped, and it’s not a consolation prize.


5. Admin modernization (the stuff nobody talks about)

Underneath the DataViews work, 7.0 ships some broader admin housekeeping that matters more than it sounds.

Design tokens now exist for colors, spacing, and typography across the admin. If you build plugins with any kind of UI, this means your interface can use the same variables as core and look consistent automatically. It also means future admin theme changes propagate correctly instead of everything breaking.

Form elements are unified. Buttons, inputs, checkboxes — consistent across the admin now. The old admin had years of accumulated visual debt from different eras of UI work. Small thing, visible improvement.

The command palette is the addition I didn’t expect to care about. Keyboard shortcut, start typing, jump to any screen or trigger any action. No menu hunting. Once you use it for a few days, the sidebar navigation feels slow in comparison. It’s the kind of thing VS Code and Linear users will immediately recognize.


6. PHP 7.4 minimum

The boring but important part: WordPress 7.0 drops PHP 7.2 and 7.3. Those versions have been out of security support for years. Most managed hosts are on PHP 8.x already, so this shouldn’t catch anyone off guard.

If you’re building plugins or themes, test against PHP 8.1 and 8.2. That’s where things are going.


Should you update?

Personal blog or content site: yes. DataViews alone is worth it, and the block improvements add up.

Client sites or WooCommerce installs: test on staging first. The DataViews change can affect admin plugins that hook into the old list table architecture. Verify compatibility before pushing to production.

Plugin and theme developers: read the Field Guide on Make WordPress Core before touching anything. There are API changes that matter, especially if your work touches the admin or anything AI-adjacent.


One last thing: waiting on the WordPress MCP

7.0 makes me more optimistic about WordPress than I’ve been in a while. After everything that happened in 2025, shipping a release this coherent is meaningful.

The thing I’m actually watching now is the MCP connector space. I’m using NovaMirror to connect WordPress to my Claude workflow, and it handles the job. But what I want — and what I think is coming — is a first-party WordPress MCP. The kind where you can talk to your site directly inside Claude, see live post data, trigger actions, review drafts, without any context switching.

That doesn’t fully exist yet. But WordPress 7.0 shipping a native AI client and Abilities API is the infrastructure that makes it possible. When an official WordPress MCP lands, that’s going to change how a lot of people work with content sites. I’ll be first in line.

NovaMirror does the job for now. But the official version is going to be something else entirely.


WordPress 7.0 was released May 20, 2026. Back up before upgrading, test on staging, then update core before plugins.