This page answers one question - which FiveM frameworks do servers actually run? - with numbers that are never hand-edited. The table reads directly from the Heartbeat.re census of public FiveM servers and re-reads it on a six-hour cycle. There is no "as of last September" on this page: the timestamp beside the table is when the census was last read.
| Framework | Share of servers | Share of players |
|---|---|---|
| ESXmost used | 45.7% | 59.3% |
| QBCore | 32.7% | 24.5% |
| Qbox | 21.6% | 16.1% |
Measured by Heartbeat.re across public FiveM servers; a server's framework is identified by the core resource it runs. Network totals were unreadable at this read, so shares are among the three major frameworks only. Shares refresh with the census - this table is never hand-edited.
How this is measured
Heartbeat.re runs a continuous census of public FiveM servers: which servers are online, how many players they hold, and which resources they run. A server's framework is identified by the framework core resource it runs - es_extended for ESX, qb-core for QBCore, qbx_core for Qbox - and the shares are those core counts against the census's network totals. The two columns:
- Share of servers - the percentage of censused servers running each framework's core.
- Share of players - the percentage of live players who were on those servers at census time.
The two can tell different stories: a framework popular with large servers can hold a bigger share of players than of servers, and the other way around. The Other row is the remainder - censused servers running none of the three major cores (standalone stacks, custom frameworks, and anything the census sees but we don't classify).
What these numbers can and cannot tell you
We publish data the way we'd want to read it - with its limits attached:
- The census covers public servers. Private or unlisted servers that the census cannot see are not counted, in any column.
- Attribution is by core resource. A renamed or heavily customized core won't be attributed and lands in Other; a server somehow running more than one core is counted for each.
- It's a snapshot. Player share especially moves with the time of day and the weekend. Watch the trend, not the decimals.
- Share is not quality. The most-used framework is not automatically the best one for your server - it's evidence of ecosystem size, nothing more. Our actual recommendation, with reasoning, lives in the framework comparison.
The three frameworks, briefly
ESX is the oldest of the three and has historically held the largest share by raw server count, with the largest legacy script ecosystem to match. It also ships the least out of the box - no modern inventory, target, fuel or phone systems by default.
QBCore is the most complete framework on its own terms: inventory, target, fuel and keys, weather, phone and more, all in-house. It recorded the weakest idle performance in our September 2025 profiler test.
Qbox is a fork of QBCore that replaces the in-house stack with high-quality external dependencies (ox_inventory, ox_fuel, NPWD and friends) while keeping QBCore resource compatibility. It has been the fastest-growing of the three. The head-to-head with its parent is in Qbox vs QBCore.
Changelog
Dated commentary lands here as the census accumulates history, so this page becomes a public record of how the framework landscape moved - not just where it stands.
- August 28, 2026 - tracker launched, reading live from the Heartbeat.re census.
Frequently Asked Questions
What is the most used FiveM framework?
The live table at the top of this page is the current answer - it refreshes automatically from the Heartbeat.re census. Historically, ESX has led by raw server count, with QBCore second and Qbox growing fastest.
How often do these numbers update?
The census itself runs continuously; this page re-reads it on a six-hour cycle, and the timestamp above the table shows exactly when the current numbers were pulled.
Where do the numbers come from?
From Heartbeat.re, which censuses public FiveM servers and the resources they run. We count the servers running each framework's core resource (es_extended, qb-core, qbx_core) against the census's network totals - the full methodology and its limits are described above, and the table is never edited by hand.
Should market share decide which framework I use?
No - it should inform, not decide. Share tells you how much third-party support and community knowledge exists for a framework; it says nothing about whether that framework fits your server. Our reasoning-first pick is in Best FiveM framework in 2026.
Building on this data? You're welcome to cite this page - it exists to be the current, sourced answer. If you're choosing a framework for a new server, start with the comparison; every script we make runs on all three.