Best Laravel Hosting in 2026: Score Hosts on Four Things, Not PHP
“It runs PHP” is not a Laravel hosting requirement. Every shared host runs PHP, and almost none of them will run a Laravel application properly.
Score hosts on four things instead — all four confirmed by Laravel’s own documentation:
- A supervisor for queue workers. Laravel’s deployment docs describe queue workers, Reverb and Octane as “long-running services” that must be restarted on deploy. Long-running is the operative word — shared hosting reaps exactly those processes.
-
A working cron. The scheduler needs one entry firing
schedule:runevery minute. - Redis. Laravel’s queue documentation is built around it, and Horizon — the queue dashboard — is Redis-only.
- Somewhere to run Octane, if you want it. More on why that changes everything below.
The app itself is trivial
We built a stock Laravel 13.25 project on PHP 8.3 and measured it:
| What’s running | Memory |
|---|---|
| Laravel serving requests | 58.7 MB |
+ one queue:work worker |
89.7 MB |
A queue worker costs about 31MB. So a Laravel app with a worker fits in 90MB, and on a 4GB box that’s roughly 2% of your memory.
This is the same finding as our Django measurements: the framework is not what you’re paying for. Your bill is decided by the database, Redis, and how many long-running processes you need — not by the application.
The exception is Magento, where the dependency stack genuinely is the cost — Adobe’s current requirements no longer list Elasticsearch, MySQL or Redis.
The scheduler, for what it’s worth, just works. We ran php artisan schedule:run and it reported no due commands, as expected. It’s a cron entry, not a hosting feature — but it’s a cron entry shared plans often don’t give you at one-minute resolution.
Why Octane changes the hosting question
From Laravel’s Octane documentation:
“Octane boots your application once, keeps it in memory, and then feeds it requests at supersonic speeds.”
Read what that means for hosting. Traditional PHP is the shared-hosting model’s best friend: a request arrives, PHP boots, responds, and dies. Octane inverts it. Your application becomes a resident daemon, like a Node or Python service.
So Octane isn’t a performance tweak you bolt on — it’s a different deployment model, and it disqualifies any host that can’t run a persistent process. Laravel supports FrankenPHP (a PHP application server written in Go), Swoole and RoadRunner, and its deployment docs note FrankenPHP can serve Laravel directly with frankenphp php-server -r public/.
The practical rule: if you might want Octane later, buy a VPS now. You cannot add it to a shared plan.
We did not measure Octane’s memory footprint — a resident application server holds your app in memory permanently, so expect it to be higher than the 58.7MB above, and to scale with worker count rather than request volume.
The two honest routes, with real totals
Route one: Forge on your own server. Laravel Forge provisions and manages the stack — its docs say Forge “installs and manages all of the tools needed to build robust Laravel applications, such as Nginx, MySQL, Redis, Memcached” — on DigitalOcean, Linode, AWS and others.
Forge’s pricing, read live:
| Plan | Cost |
|---|---|
| Hobby | $12/mo — unlimited Laravel VPS servers, 1 external server |
| Growth (recommended) | $19/mo — 5% discount on Laravel VPS |
| Business | $39/mo — 15% discount, server monitoring, dedicated support |
So the Forge-on-DigitalOcean total is $12 + $24 = $36/month for a 4GB droplet with the whole stack configured, queue workers supervised and deploys automated. You own the server; Forge owns the tedium.
Route two: managed hosting. Cheaper on paper — Cloudways starts around $11 for 2GB — but read the support scope before assuming it covers you. Cloudways publishes exclusions on every tier including debugging custom code and application-level security issues. On a Laravel app, “custom code” is your entire application. We went through what that boundary actually means.
The honest comparison: Forge at $36 automates your server. Managed hosting at $11–24 manages the server and explicitly not your app. Neither one debugs your queue job.
Where shared hosting fails, concretely
No supervisor. You can’t keep queue:work alive. Some people work around this with a cron running queue:work --once, which functions but is not a queue system.
Cron at the wrong resolution. Laravel’s scheduler assumes a per-minute cron. Plans offering five- or fifteen-minute minimums break everyMinute() silently.
No Redis. You’ll fall back to the database queue driver, which works and puts your queue load onto the database you’re already straining.
No Octane, ever. Covered above.
The one-line test: if you can’t SSH in and run something under systemd or supervisor that survives logout, you can’t run Laravel properly — the same boundary that rules out Docker on shared plans.
The hosts
Cloudways — from $11 for 2GB, billed monthly, managed on top of DigitalOcean or AWS, with Redis and supervisor available. The lowest-effort route if you want someone else patching the server, provided you accept that your application code is explicitly out of its support scope.
DigitalOcean — $24 for 4GB, the canonical Forge target and the best documentation in the category. The $36/month Forge combination is the default recommendation for most serious Laravel teams, and per-second billing since January 2026 makes staging servers cheap.
Hostinger — $11.99 renewal for 4GB, $14.99 for 8GB, root access and one-click Docker templates. The cheapest VPS that will genuinely run the full stack, and its refund terms cover renewals. Buy 8GB if MySQL and Redis share the box.
Hosting.com (formerly A2 Hosting) — long-standing PHP support and a familiar control-panel environment. Confirm root access and supervisor availability, not just that Laravel is “supported” — we have not verified its current plan structure.
ScalaHosting — managed VPS with its own SPanel instead of cPanel, removing a licence cost from the stack. A reasonable middle if you want managed infrastructure and still need to run daemons.
Liquid Web — the clearest published support boundary of any host here, with a tier matrix stating exactly which level stops troubleshooting your server. Its fully managed tier is the answer if you need someone else responsible for the OS, at a different price class from everything above.
Hetzner — the most RAM and cores per euro, which matters once MySQL, Redis, Horizon and several queue workers share a box. Forge supports custom servers. Two caveats: availability has been limited since June and backups exclude attached Volumes — relevant if MySQL lives on a Volume, which it should.
What we’d buy
An 8GB VPS with Forge on top, running Nginx, MySQL, Redis, Horizon and your workers on one box. Your app’s 90MB leaves the rest for the database, and the $12 Forge tier removes the work that makes people pay for managed hosting in the first place.
Split the database out when recovery matters more than money, not before.
How we checked this
The memory figures are ours, measured on 18 August 2026. We built a container with PHP 8.3.33 and created a stock project with composer create-project laravel/laravel, which installed Laravel Framework 13.25.0. We served it with php artisan serve, sent 30 requests, then started a queue:work worker in the same container and re-measured with docker stats. We also ran php artisan schedule:run, which correctly reported no due commands.
Read those numbers as a floor, not a forecast. A stock Laravel project has no packages, no models, no Eloquent result sets and no application code. A real app will use materially more, and artisan serve is PHP’s development server rather than the php-fpm-and-Nginx arrangement you’d deploy. What the measurement establishes is the shape — the framework baseline is small and a queue worker is roughly 31MB — not what your production server needs. Measure your own app the same way.
We did not measure Octane, FrankenPHP, Swoole or RoadRunner at all, so there are no figures for them here. The description of what Octane does is quoted from Laravel’s documentation, not observed.
The Laravel platform requirements are quoted from Laravel’s official docs, read the same day: the deployment page’s description of queue workers, Reverb and Octane as long-running services needing restart on deploy; the FrankenPHP serving instructions; the Octane page’s “boots your application once, keeps it in memory” description; the queue documentation’s Redis configuration and Horizon’s Redis requirement; and the Forge description of the tools it installs.
The Forge pricing was read live from forge.laravel.com — $12 Hobby, $19 Growth and $39 Business per month. The $36/month total is our arithmetic on Forge Hobby plus DigitalOcean’s $24 4GB droplet.
The big gap: this article’s brief asked us to deploy the same Laravel app with a queued job and a scheduled command to each host and report what broke. We did not do that — we don’t hold accounts at these seven providers, and a deployment matrix is only worth publishing if every cell was actually run. So the host section below ranks on product class and published capability, not on a deployment we performed. That’s a real weakness on a page meant to help you choose, and the four-point checklist exists so you can run the test yourself.
We have not verified Hosting.com’s current Laravel plan structure, which is why our note on it is a caution. Provider prices other than Forge’s and DigitalOcean’s are figures established across this series rather than re-verified today. The Cloudways support exclusions are from its published support page as verified earlier in this series.
What we did not do: deploy Laravel to any host, run Horizon, configure supervisor on a real server, use Forge, or benchmark anything.
The host links above are affiliate links, and Laravel Forge — the product we recommend most strongly — is not one of them.
FAQ
What’s the best hosting for Laravel?
A VPS with root access, because Laravel needs supervised queue workers, a per-minute cron and Redis. Forge on a $24 DigitalOcean droplet at $12/month is the default for serious apps — $36 total.
Can I run Laravel on shared hosting?
The web requests, often yes. The rest, usually not: no supervisor for queue workers, cron frequently limited to five- or fifteen-minute intervals, no Redis, and no Octane.
How much RAM does Laravel need?
We measured a stock Laravel 13 project at 58.7MB serving requests and 89.7MB with a queue worker. Real apps use more, but the framework isn’t your constraint — the database usually is.
What is Laravel Octane and do I need it?
It boots your app once and keeps it in memory, serving requests from a resident process. You don’t need it, but wanting it later means buying a VPS now — it can’t run on shared hosting.
How much does Laravel Forge cost?
$12/month for Hobby with unlimited Laravel VPS servers, $19 for Growth and $39 for Business. That’s on top of whatever the server costs.
Is managed Laravel hosting worth it over Forge?
Managed hosting manages the server, and typically excludes your application code from support. Forge automates the server and leaves you in control. Neither will debug your queue job.
Do I need Redis for Laravel queues?
Not strictly — the database driver works. But Horizon requires Redis, and the database driver puts queue load onto the database you’re already using.
