Does anybody still use Ruby on Rails? Yes, by every measure that can be checked: the companies running it
in production, the releases shipping on schedule, and the money being spent on the framework's upkeep.
The question gets asked anyway, and it deserves better than a logo wall assembled in 2015, so everything
below is dated, named and sourced, including the part where Rails genuinely lost. If the prior question is
what the framework even does, Ruby on Rails, answered covers that.
Who runs Rails in production right now
GitHub is a Ruby on Rails monolith and has been since the beginning. GitHub's own engineering blog puts
the application at nearly two million lines of code, worked on by more than a thousand engineers a day and
deployed as often as twenty times daily. It also describes a scheduled workflow that opens a pull request
every Monday bumping GitHub's Rails version to the latest commit on the framework's main branch. A company
that upgrades its framework weekly is not nursing a legacy system, it is living on the framework's edge on
purpose.
Shopify runs the other famous one. The Rails 8.1 announcement says that Shopify and HEY had been running
it in production for months before the release was cut, which is how Rails releases have worked for years:
the largest applications run the unreleased version so that the release is already proven on the day it
ships.
37signals, where Rails was extracted from in the first place, still writes it and still deploys it. The
Rails 8 announcement put a number on that in passing: Solid Queue was already running twenty million jobs
a day for HEY alone before it became a framework default.
Two companies have published proper case studies with the Rails Foundation, which is the most detailed
evidence available. Doximity has been on Rails for fifteen years, with more than 150 developers pushing to
production over eighty times a day on a monolith they upgraded from Rails 2.3.5. Cookpad moved from
ColdFusion to Rails in 2007 and now serves more than 100 million home cooks a month on it.
The companies that use Ruby on Rails and pay for it
Paying is a better signal than a logo, and the Rails Foundation publishes who pays. Members fund the
framework's documentation, education, marketing and events rather than merely consuming them. Core members
are Cookpad, Doximity, Fin, Fleetio, GitHub, Judge.me, Procore, Shopify, 1Password and 37signals.
Contributing members include AppSignal, Chime, Clio, Fullscript, Gusto, Planning Center, Saeloun, SerpApi
and TableCheck.
Seven companies joined during 2025, and that number matters more than the total: a framework on its way
out does not acquire new corporate funders. Judge.me moved up to core membership in the same year.
What the membership list does not prove is how much Rails each of those companies runs. Funding the
foundation demonstrates a commercial interest in the framework's future, not that every one of those
products is a Rails monolith, and this page will not claim the second thing on evidence for the first. For
the five companies in the section above, the evidence is direct and published.
What has actually shipped since 2024
Rails 8.0 shipped on 7 November 2024, and its headline was removing infrastructure rather than adding
features. Solid Queue, Solid Cache and Solid Cable moved background jobs, caching and WebSockets into the
database, which is what lets a new Rails application run with no Redis at all. Propshaft replaced Sprockets
as the default asset pipeline. Kamal 2 arrived preconfigured, with a proxy called Thruster in front of
Puma, under the banner "no PaaS required". And bin/rails generate authentication began generating a
session-based, password-resettable authentication system with a Session model, a SessionsController, a
PasswordsMailer and an Authentication concern, instead of sending every new application to a gem.
Rails 8.1 shipped on 22 October 2025, described in its own announcement as the work of over 500
contributors across 2500 commits. In it: Active Job continuations, which break a long-running job into
discrete steps so that a restart resumes from the last completed step instead of the beginning; structured
event reporting; a local CI runner; native Markdown rendering; command-line credentials fetching;
deprecated associations; and registry-free Kamal deployments.
Maintenance has continued on schedule since. Versions 8.0.5 and 8.1.3 were released on 24 March 2026, and
coordinated security releases for the 7.2, 8.0 and 8.1 lines went out on 29 July 2026. Three supported
release lines getting security patches on the same day is the least glamorous evidence on this page and
close to the most convincing. Every claim here comes from the announcements and guides published on
rubyonrails.org, and which of the official sources is the current one is
its own question worth answering before you trust a tutorial.
Where Rails genuinely lost ground
Measured usage is where the loss is visible. In the 2025 Stack Overflow Developer Survey, 5.9% of all
respondents reported working with Ruby on Rails, against 49.1% for Node.js and 46.9% for React. The Rails
figure has barely moved in three years, sitting between 4.7% and 5.9% across the 2023, 2024 and 2025
surveys, and that flatness is the actual finding: Rails did not collapse, it was overtaken while standing
still.
Territory has been lost in specific places, and they are worth naming rather than waving away. Machine
learning never came to Ruby, so the libraries a team needs to train or serve a model are written for
Python, and a product whose core is a model does not start in Rails. Anything that ships a native mobile
client first also gets less out of Rails than a web application does, because Hotwire, the strongest
argument for keeping everything inside one Rails codebase, does nothing for an iOS app that wants a JSON
API and would have been just as happy with any other backend.
Rails also lost the default. A team starting a web product in 2015 had to justify not using Rails, and a
team starting one now has to justify using it. Nothing about the framework got worse in between, but the
burden of proof moved, and that is a real loss rather than a perception problem.
Ruby on Rails vs JavaScript is where that specific decision gets worked
through properly.
What the people saying Rails is dead are right about
Critics have a real case, and the honest version of it is not about production use at all. The question
behind this one usually gets typed as "is Ruby on Rails dead", and what people mean by it is closer to
"will I still be hiring for this in five years" and "will the next developer I hire already know it". Those
are fair worries, and none of the release dates above answer them.
Performance is the second fair point. Ruby 4.0, released on 25 December 2025, ships a new just-in-time
compiler called ZJIT, and the team that built it wrote on Shopify's Rails at Scale blog that it is faster
than the interpreter but not yet as fast as YJIT, and that people should hold off deploying it to
production for now. A language still working its JIT story out in public in 2026 is a language that started
slow, and pretending otherwise would insult a reader who can benchmark.
Mindshare among people learning to build for the web is the third. Rails is not the framework most new
developers meet first any more, and over a decade that changes who is available to hire and what they
expect a codebase to look like. Is Ruby on Rails hard to learn covers what
picking it up actually costs today, which is the part of this worry you can do something about.
Rails World 2026 opens in Austin this week
Rails World 2026 runs on 23 and 24 September at the Palmer Center in Austin, with the Rails at Scale Summit
and the workshops the day before. The 2025 edition in Amsterdam drew 825 attendees, and the Austin venue is
set up for 1,200. Conference size is a weak signal taken alone, and it earns its place here only because it
is checkable: the dates, the venue and the capacity are all published, unlike the general impressions that
usually carry this argument in either direction.
The verdict, and what would change it
Ruby on Rails in 2026 is a safe bet for a web product where logged-in users change data, and the case for
that is not sentiment. The largest applications on Rails upgrade weekly rather than freeze. Two major
releases in two years removed dependencies instead of adding them. The companies paying for the
framework's upkeep are still joining rather than leaving. Against that, the only counter-evidence anybody
produces is a survey percentage that has not fallen.
Looking at a Rails application beats reading about one, and
a Ruby on Rails example application walks through a real one feature by
feature, which is a faster way to decide than any argument on this page.
What would overturn the verdict, stated so that you can hold it to account: a core member of the Rails
Foundation publicly moving its main application off Rails, or two consecutive years passing with no minor
release. Neither has happened. A falling survey share would not overturn it on its own, because that share
has been flat for three years while the framework itself got materially better, and a number that tracks
nothing is not evidence of anything.
Does anybody still use Ruby on Rails? Yes, by every measure that can be checked: the companies running it in production, the releases shipping on schedule, and the money being spent on the framework's upkeep. The question gets asked anyway, and it deserves better than a logo wall assembled in 2015, so everything below is dated, named and sourced, including the part where Rails genuinely lost. If the prior question is what the framework even does, Ruby on Rails, answered covers that.
Who runs Rails in production right now
GitHub is a Ruby on Rails monolith and has been since the beginning. GitHub's own engineering blog puts the application at nearly two million lines of code, worked on by more than a thousand engineers a day and deployed as often as twenty times daily. It also describes a scheduled workflow that opens a pull request every Monday bumping GitHub's Rails version to the latest commit on the framework's main branch. A company that upgrades its framework weekly is not nursing a legacy system, it is living on the framework's edge on purpose.
Shopify runs the other famous one. The Rails 8.1 announcement says that Shopify and HEY had been running it in production for months before the release was cut, which is how Rails releases have worked for years: the largest applications run the unreleased version so that the release is already proven on the day it ships.
37signals, where Rails was extracted from in the first place, still writes it and still deploys it. The Rails 8 announcement put a number on that in passing: Solid Queue was already running twenty million jobs a day for HEY alone before it became a framework default.
Two companies have published proper case studies with the Rails Foundation, which is the most detailed evidence available. Doximity has been on Rails for fifteen years, with more than 150 developers pushing to production over eighty times a day on a monolith they upgraded from Rails 2.3.5. Cookpad moved from ColdFusion to Rails in 2007 and now serves more than 100 million home cooks a month on it.
The companies that use Ruby on Rails and pay for it
Paying is a better signal than a logo, and the Rails Foundation publishes who pays. Members fund the framework's documentation, education, marketing and events rather than merely consuming them. Core members are Cookpad, Doximity, Fin, Fleetio, GitHub, Judge.me, Procore, Shopify, 1Password and 37signals. Contributing members include AppSignal, Chime, Clio, Fullscript, Gusto, Planning Center, Saeloun, SerpApi and TableCheck.
Seven companies joined during 2025, and that number matters more than the total: a framework on its way out does not acquire new corporate funders. Judge.me moved up to core membership in the same year.
What the membership list does not prove is how much Rails each of those companies runs. Funding the foundation demonstrates a commercial interest in the framework's future, not that every one of those products is a Rails monolith, and this page will not claim the second thing on evidence for the first. For the five companies in the section above, the evidence is direct and published.
What has actually shipped since 2024
Rails 8.0 shipped on 7 November 2024, and its headline was removing infrastructure rather than adding features. Solid Queue, Solid Cache and Solid Cable moved background jobs, caching and WebSockets into the database, which is what lets a new Rails application run with no Redis at all. Propshaft replaced Sprockets as the default asset pipeline. Kamal 2 arrived preconfigured, with a proxy called Thruster in front of Puma, under the banner "no PaaS required". And
bin/rails generate authenticationbegan generating a session-based, password-resettable authentication system with aSessionmodel, aSessionsController, aPasswordsMailerand anAuthenticationconcern, instead of sending every new application to a gem.Rails 8.1 shipped on 22 October 2025, described in its own announcement as the work of over 500 contributors across 2500 commits. In it: Active Job continuations, which break a long-running job into discrete steps so that a restart resumes from the last completed step instead of the beginning; structured event reporting; a local CI runner; native Markdown rendering; command-line credentials fetching; deprecated associations; and registry-free Kamal deployments.
Maintenance has continued on schedule since. Versions 8.0.5 and 8.1.3 were released on 24 March 2026, and coordinated security releases for the 7.2, 8.0 and 8.1 lines went out on 29 July 2026. Three supported release lines getting security patches on the same day is the least glamorous evidence on this page and close to the most convincing. Every claim here comes from the announcements and guides published on rubyonrails.org, and which of the official sources is the current one is its own question worth answering before you trust a tutorial.
Where Rails genuinely lost ground
Measured usage is where the loss is visible. In the 2025 Stack Overflow Developer Survey, 5.9% of all respondents reported working with Ruby on Rails, against 49.1% for Node.js and 46.9% for React. The Rails figure has barely moved in three years, sitting between 4.7% and 5.9% across the 2023, 2024 and 2025 surveys, and that flatness is the actual finding: Rails did not collapse, it was overtaken while standing still.
Territory has been lost in specific places, and they are worth naming rather than waving away. Machine learning never came to Ruby, so the libraries a team needs to train or serve a model are written for Python, and a product whose core is a model does not start in Rails. Anything that ships a native mobile client first also gets less out of Rails than a web application does, because Hotwire, the strongest argument for keeping everything inside one Rails codebase, does nothing for an iOS app that wants a JSON API and would have been just as happy with any other backend.
Rails also lost the default. A team starting a web product in 2015 had to justify not using Rails, and a team starting one now has to justify using it. Nothing about the framework got worse in between, but the burden of proof moved, and that is a real loss rather than a perception problem. Ruby on Rails vs JavaScript is where that specific decision gets worked through properly.
What the people saying Rails is dead are right about
Critics have a real case, and the honest version of it is not about production use at all. The question behind this one usually gets typed as "is Ruby on Rails dead", and what people mean by it is closer to "will I still be hiring for this in five years" and "will the next developer I hire already know it". Those are fair worries, and none of the release dates above answer them.
Performance is the second fair point. Ruby 4.0, released on 25 December 2025, ships a new just-in-time compiler called ZJIT, and the team that built it wrote on Shopify's Rails at Scale blog that it is faster than the interpreter but not yet as fast as YJIT, and that people should hold off deploying it to production for now. A language still working its JIT story out in public in 2026 is a language that started slow, and pretending otherwise would insult a reader who can benchmark.
Mindshare among people learning to build for the web is the third. Rails is not the framework most new developers meet first any more, and over a decade that changes who is available to hire and what they expect a codebase to look like. Is Ruby on Rails hard to learn covers what picking it up actually costs today, which is the part of this worry you can do something about.
Rails World 2026 opens in Austin this week
Rails World 2026 runs on 23 and 24 September at the Palmer Center in Austin, with the Rails at Scale Summit and the workshops the day before. The 2025 edition in Amsterdam drew 825 attendees, and the Austin venue is set up for 1,200. Conference size is a weak signal taken alone, and it earns its place here only because it is checkable: the dates, the venue and the capacity are all published, unlike the general impressions that usually carry this argument in either direction.
The verdict, and what would change it
Ruby on Rails in 2026 is a safe bet for a web product where logged-in users change data, and the case for that is not sentiment. The largest applications on Rails upgrade weekly rather than freeze. Two major releases in two years removed dependencies instead of adding them. The companies paying for the framework's upkeep are still joining rather than leaving. Against that, the only counter-evidence anybody produces is a survey percentage that has not fallen.
Looking at a Rails application beats reading about one, and a Ruby on Rails example application walks through a real one feature by feature, which is a faster way to decide than any argument on this page.
What would overturn the verdict, stated so that you can hold it to account: a core member of the Rails Foundation publicly moving its main application off Rails, or two consecutive years passing with no minor release. Neither has happened. A falling survey share would not overturn it on its own, because that share has been flat for three years while the framework itself got materially better, and a number that tracks nothing is not evidence of anything.