Laravel 13 (2026): every new feature explained, and what it means for your web app
Laravel 13 shipped a first-party AI SDK, native vector search, PHP attributes and team-based multi-tenancy. Here is what actually changed in March 2026 — and why it makes custom web apps faster and cheaper to build.
F
Fepiq Team
Fepiq
Taylor Otwell shipped Laravel 13 on March 17, 2026, and — unlike most major version bumps — this one is not just a dependency refresh. It ships a first-party AI toolkit, native vector search, a wave of PHP attributes, and team-based multi-tenancy baked into the starter kits. If your business runs on Laravel, or you are choosing a stack for your next web app, this release changes the calculation.
We build on Laravel for every client project, so we read the release notes the way an accountant reads a tax law change: what does this let us ship faster, and what does it let us stop building by hand? Here is the honest breakdown.
Scattered queue config and wasteful cache round-trips
Requirements moved too: Laravel 13 needs PHP 8.3 or higher and drops PHP 8.2. Bug fixes run through Q3 2027 and security patches through Q1 2028, so this is a safe version to build a multi-year business app on.
The Laravel AI SDK: AI features without five separate packages
Until now, adding AI to a Laravel app meant installing an OpenAI client, a separate embeddings package, a vector-store adapter, and writing your own glue code for tool-calling. Laravel 13 ships a first-party AI SDK with one unified API for text generation, tool-calling agents, embeddings, audio, images, and vector-store integrations.
For a business owner, the practical effect is speed. A feature like "let customers ask a question and get an answer grounded in our product catalogue" used to be a multi-week integration project. With a first-party SDK maintained by the framework itself, it becomes a feature we can scope in days, not sprints — and one that will not break the next time a third-party package changes its API.
Vector search, built into the query builder
Semantic search — "find things like this" instead of "find things that match this exact word" — used to require standing up a separate vector database and keeping it in sync with your primary one. Laravel 13 adds native vector query support for PostgreSQL with pgvector, and MariaDB now supports creating vector indexes directly through Laravel's schema builder in a normal migration.
That matters for very ordinary business features: a support inbox that surfaces similar past tickets, a product search that understands "cheap gold ring" without the word "cheap" appearing in any title, a documents system that finds the right clause across thousands of contracts. Before this release, all three needed a second database. Now they are a migration and a query.
PHP attributes: less boilerplate, faster code review
Laravel has been steadily moving configuration out of class properties and arrays and into PHP attributes — the "#[...]" syntax that sits directly above the method or class it configures. Laravel 13 expands this across more than 15 places in the framework, including controllers, jobs, and console commands.
#[Middleware] and #[Authorize] on controller methods, instead of separate middleware arrays and policy lookups.
#[Tries], #[Backoff], #[Timeout], and #[FailOnTimeout] on queued jobs, replacing scattered public properties.
Attribute-based configuration for models and console commands, colocated with the code it affects.
None of this is a breaking change — the old syntax still works. But for teams reviewing pull requests, attributes mean the rule and the code it governs sit on the same screen, which is a small thing that adds up to fewer review cycles and fewer "wait, where is this job's retry count set" conversations.
Smaller wins that add up
Team-based multi-tenancy is back in the official starter kits, built on lessons from Jetstream's original Teams feature — useful the moment your product needs "workspaces" or "branches" rather than one account per business.
First-party JSON:API resource classes handle spec-compliant serialization, relationships, sparse fieldsets, and headers automatically — relevant if you are exposing data to a mobile app or a partner integration.
Queue::route() lets you define which queue and connection a job class uses in one place, instead of hunting through dispatch calls to find where a job actually runs.
Cache::touch() extends a cached item's expiry without re-fetching or re-storing the value — a small efficiency win that matters at scale (a single EXPIRE in Redis instead of a get-then-put round trip).
What this means if you are not the developer
If you are a founder or operator deciding what to build your next web app on, the honest translation of all of the above is: things that used to be custom engineering work are now framework features. AI-assisted search, semantic matching, multi-tenant workspaces, and spec-compliant APIs used to be line items that made a quote bigger. On Laravel 13, several of them are close to "turn it on."
“Every feature the framework ships for free is a feature your quote does not have to pay for.”
That is also why the gap between a "simple website" and a "real business web app" keeps shrinking in your favour. Two years ago, a searchable admin dashboard with AI-assisted lookup was an enterprise-grade project. Today, on Laravel 13, it is a well-scoped few weeks of work for a small, disciplined team.
Should you upgrade now?
For an existing Laravel 11 or 12 application, the team has explicitly prioritised minimizing breaking changes, so most apps can upgrade in well under a day of engineering time. Confirm the practical checklist before you do:
Check your hosting environment supports PHP 8.3 or higher — this is the one hard requirement.
Audit any third-party packages for Laravel 13 compatibility before upgrading a production app.
Run your test suite against the new version in staging before touching production.
If you are starting a new build from scratch, there is no reason to start anywhere but Laravel 13.
Building a new web app, or thinking about what an upgrade unlocks for yours? Let us scope it.
The headline additions are a first-party Laravel AI SDK, native vector/semantic search support for PostgreSQL and MariaDB, expanded PHP attributes across controllers, jobs and console commands, team-based multi-tenancy in the starter kits, first-party JSON:API resources, and smaller quality-of-life additions like Queue::route() and Cache::touch(). It requires PHP 8.3 or higher.
Is Laravel 13 stable enough for a production business app?
Yes. Laravel 13 shipped with bug fixes committed through Q3 2027 and security fixes through Q1 2028, following the same support policy as every other major Laravel release. It is a safe foundation for a multi-year business application.
Does Laravel 13 support AI features out of the box?
Yes, via the new first-party Laravel AI SDK, which provides one API for text generation, tool-calling agents, embeddings, audio, images, and vector-store integrations — replacing the need to wire together several separate third-party packages.
Does upgrading to Laravel 13 make my project cheaper to build?
Indirectly, yes. Features that used to require custom engineering — semantic search, multi-tenant workspaces, spec-compliant APIs — are now closer to framework defaults. That does not make the framework free, but it does shrink the amount of custom code a team has to write, test and maintain to ship the same feature.
Fepiq builds custom web apps, dashboards and AI-assisted features on Laravel for Indian businesses — from a first version in weeks to an app that keeps evolving with your business. If you are scoping a build on Laravel 13, or wondering whether an existing app is worth upgrading, message us on WhatsApp and we will tell you honestly.
Get new posts in your inbox
Occasional, no-fluff notes on shipping modern software — startups, automation, Laravel, Shopify and more. No spam, unsubscribe anytime.
Laravel Reverb or a Node.js microservice for real-time SaaS features? A 2026 decision framework covering scaling, cost, and the hybrid pattern most teams ship.
RDS PostgreSQL vs. Aurora PostgreSQL for SaaS in 2026: real cost numbers, when each wins, and a 5-question framework to pick the right one for your app.