There is now a formal proposal to merge the WP AI Client into core for the WordPress 7.0 Release. The documentation reads like a routine API addition, which undersells it. Anyone following the Core Dev Chat knows the decision has to happen right before Beta 1, scheduled for February 19, 2026.
I have watched core merges go sideways before; the initial REST API rollout comes to mind. The WP AI Client conversation is less about “adding AI” than about setting up provider-agnostic infrastructure, and the friction sits in the Abilities API underneath rather than in the AI part. Ticket #64596 spells out the bottleneck: core abilities can no longer be iterated on from a feature plugin without creating serious friction for early adopters.
The WordPress 7.0 release squad and timeline
The release squad is set and the bug scrubs are running. If you manage production sites, the “Release Party” sounds like marketing, but it is also your stability deadline. Beta 1 lands tomorrow, which leaves the Core team in a race condition between merging the AI Client and keeping the WordPress 7.0 Release from breaking legacy hooks.
I went through the wider implications in my Core Architect’s Critique of the AI Client. Opinion is moving toward putting the infrastructure in core, mostly so every plugin developer stops bundling their own bloated OpenAI or Anthropic SDK.
Where the Abilities API friction is
The abilities named in Ticket #64596 are the bridge between WordPress and LLMs. If they miss the WordPress 7.0 Release, developers wait until 7.1 for a stable interface. Six months of waiting to iterate on “Infrastructure as Code” is long enough to kill adoption.
Compare how AI interactions used to be handled with how the 7.0 infrastructure, assuming the merge, runs a capability check for a generative ability.
<?php
/**
* The "Naive" Way: Hardcoding provider logic in your plugin.
* This leads to technical debt and API key management nightmares.
*/
function bbioon_old_school_ai_call( $prompt ) {
$api_key = get_option( 'my_openai_key' );
$response = wp_remote_post( 'https://api.openai.com/v1/completions', [
'body' => json_encode( [ 'prompt' => $prompt ] ),
'headers' => [ 'Authorization' => 'Bearer ' . $api_key ],
] );
return json_decode( wp_remote_retrieve_body( $response ) );
}
The WordPress 7.0 Release approach moves that logic into a single client. Rather than tracking endpoints, you ask whether the site has the ability to perform a task, and it stops mattering whether a local WASM model or a remote API answers.
<?php
/**
* The WordPress 7.0 Way: Using the proposed AI Client and Abilities API.
* This is provider-agnostic and cleaner.
*/
function bbioon_modern_ai_interaction( $content ) {
// Check if the core AI Client is available and the site supports 'summarization'
if ( ! function_exists( 'wp_ai_get_client' ) ) {
return $content;
}
$client = wp_ai_get_client();
// Check for a specific 'ability' defined in Ticket #64596
if ( $client->supports_ability( 'text-summarization' ) ) {
return $client->summarize( $content );
}
return $content;
}
The decision needed for Beta 1
Core is split on this. One side wants the AI Client to stay a “canonical plugin” so it can keep iterating quickly. Felix Arntz and others want it merged into the WordPress 7.0 Release as the “developer infrastructure” the ecosystem is missing, the way the REST API was years ago. If it does not land in 7.0, Yoast, Jetpack and WooCommerce each implement AI their own way and the ecosystem fragments.
The remaining release milestones are listed in the WordPress 7.0 Roadmap.
If this WordPress 7.0 Release work is eating your dev hours, hand it over to me. I have been working with WordPress since the 4.x days.
What to watch as a developer
Whether the AI Client merges tomorrow or stays in “Experiments,” the move toward standardized abilities is happening. Watch Trac tickets #60029 and #61796. They cover the plumbing underneath, and that plumbing decides whether 7.0 is a quiet upgrade or a debugging marathon.
Core discussions get posted on the official Make WordPress Core blog. Get your testing environments in order this week, because Beta 1 is almost here.
