How's Strapi 5 in 2026?

By Tuukka Tähtinen

August 21st, 2026

Strapi 5 has now been out for a little under two years, so I thought it's time to recap some of the released updates, issues, and upcoming development.

Features of new

MCP server functionality has been added, along with a batch of other improvements targeted towards AI use in the dashboard and AI compatible documentation. You are now able to connect your AI agents to both your front-end and backend to create schemas and initial module drafts off of a design file.

Additionally, fields can now be conditionally displayed, however limited the conditions may be. Conditions are currently limited to a single condition which may depend on either a boolean or an enumeration field.

Live previews can now also be configured for an improved content editing experience (on a paid plan) to display changes inside the Strapi editor in an iframe.

Retrospection of usage

An ongoing nit of mine has been how arrays and objects are handled.

To display an array of strings, you'll need to have an object that includes a string field. For a number array, you'll again need an object with a number field.

For each unique object configuration, you’ll have to create a unique schema, which is going to end up polluting your code base, especially on an ever-expanding website. This is somewhat alleviated with conditional fields, but with more options comes more complex code.

Eventually, you will (hopefully) be able to remove some component schemas when pages get redesigned and reworked.

Especially considering that if you end up needing to nest components, or creating simple arrays, those will all need schemas. And if one of those schemas is used on multiple components but needs updates on one? Too bad they share the schema, swap the schema on one, and you’ll end up with data loss without a manual data migration first. Don’t swap the schema? You end up with complicated schemas that have fields that don't have an effect on some use cases depending on their parent.
If you can properly plan for the schema configurations when doing the designs for components, you can negate some of these issues.

Bugs of the past

It’s no secret that software development is often chaotic and can provide unexpected results such as breaking bugs or unintended behavior, which by itself is already disappointing when a supposedly stable update ends up breaking something. There have been a couple instances of issues specifically with relation fields. At one point in time, an update caused relation fields to fetch both the draft and published versions of the linked item. You also don’t want the default sort order of an array to be changed on a seemingly minor update. I'm not one to hold a grudge but you hate to see it.

Another issue that had persisted for a while but has since been fixed, was sudden logouts in the middle of editing on the dashboard. Quite annoying if you had unsaved changes that you lost without warning, but that’s water under the bridge now.

Additions of the future

Not much is listed on their public roadmap, aside from mobile display improvements and the option to group content types on the dashboard for an improved UX experience.

More complex field conditions are said to be coming soon, such as only showing a field if a relation is set and being able to set multiple required conditions.