How Many FiveM Servers Are There? We Counted 35,145

Almost every source that tells you how many FiveM servers exist hands you a single number and never says how that number was produced. We scan the Cfx.re master server list every ten minutes. Every figure below comes from one single scan, and which scan it was, what it covers and what it does not are all written down.

How this was measured

Two separate universes run through this piece and we never mix them. The scan is the set of servers standing in the Cfx.re list during one single pass — a snapshot. The registry is every server seen in that list at least once during the window — a cumulative ledger. They are not the same thing, and the gap between them is one of the more interesting findings here.

The concurrent player count deliberately comes from one scan. Summing the last known value per server produces a larger result, but that total stacks figures from different moments and is therefore correct at no moment at all. A single scan belongs to a single instant; we publish the accurate number rather than the bigger one.

One scan: 35,145 servers, 162,806 players

The Cfx.re master list carried 35,145 servers at 17:01 (TR) on 28 September 2026, holding 162,806 players between them. That averages out to 4.6 players per server.

That average describes almost no actual server, as the rest of this piece shows: in the same scan 27,327 servers — 77.8% of the list — had nobody on them at all. There is no "average FiveM server". There is a scene made of a few hundred servers and a tail made of tens of thousands of empty ones.

The registry holds 46,797 servers, but never all at once

Across the window we recorded 46,797 distinct servers. The gap between that and the 35,145 present in a single scan is 11,652 servers. It comes from restarts, from servers dropping out of the listing for a while, and from servers genuinely shutting down. Looking back one hour from the measurement, 36,206 of them had appeared in the list — meaning a slice of the ledger has been absent for hours.

Of that ledger, 17,715 servers — 37.9% — did not have a single player at any point in the window. From here on, "observed peak" means one specific thing: the highest concurrent player count we measured for that server inside the window. Not the slot count it advertises, and not its all-time peak either, because we did not measure the overnight hours.

Where do the players concentrate?

The table below comes from that same scan. One caveat first: the tag here is the language-and-region label the server declares for itself, not a geographic measurement. A server can be hosted in Germany and still carry a Turkish tag. We measure the label, not the location.

The most striking row is the first one. The 378 servers tagged Thailand hold 32,100 players — 84.9 each. In the same scan, the 5,130 servers tagged United States held just 9,831 players between them, or 1.9 each. Server counts and player counts measure two independent things, which is exactly why rankings built on server counts mislead.

Servers that report no region at all: 10,945

A large group is missing from that table. In the scan, 10,945 servers either carry the root-AQ tag or carry no tag at all. root-AQ is not a region: it is Antarctica's country code, and it is the bucket a server falls into when it sets no region. We removed these from the regional table, because leaving them in would invent a FiveM scene that does not exist.

That exclusion is why the table totals do not add up to the scan totals. The 24,200 servers that do report a region hold 157,903 players; the remaining 10,945 hold 4,903. So 31.1% of the scanned servers report no region, yet they account for only 3.0% of the players. Leaving the region unset is largely a property of small and empty servers. Among those that do report one we counted 623 distinct tags.

Turkey ranks 9th by player count

Servers tagged Turkish hold 6,185 players in the scan, ranking 9th among the 623 region tags. By server count they rank 6th with 683 servers. That is 2.8% of the region-reporting servers but 3.9% of the region-reporting players. Turkey carries a larger share of players than of servers.

We keep an up-to-date list of the servers carrying that tag, with a page for each one: the live list of Turkish servers

176 servers hold nearly half of all players

Splitting that same scan into buckets by player count gives the following shape. The bars show players; the figure in brackets is how many servers sit in that bucket.

176 servers had 200 or more players. That is 0.50% of the scanned servers and 49.6% of the scanned players. Drop the threshold to 50 players and 671 servers — 1.91% of the list — account for 78.1% of everyone playing.

At the other end, 27,327 servers were completely empty and 6,097 had 9 players or fewer. That is the shape of the FiveM ecosystem: a narrow head and a very long tail. Anyone opening a new server finds out which end they landed on within the first week.

Every server in this scan is listed here with search and region filters, and each one has its own measurement page: the server directory

What we did not measure

• The night. Our uninterrupted coverage runs from 09:11 to 17:01, with four gaps longer than twenty minutes (the largest half an hour). That is why this piece contains no "daily peak", no "busiest hour" and no time-of-day profile.
• The real peak of the day. Player counts were still climbing in our last scans; the highest value we recorded is not a peak, it is simply where our window ends.
• Weekday versus weekend, monthly growth, and the rate at which servers open and close. All three need at least two weeks of unbroken series, which this window does not contain.
• The inside of the servers: which framework and which resources they run is not measured here. That is a separate exercise.
• Whether the region tag is accurate. It is self-declared, and we did not check it against where the server is actually hosted.
• The players. Everything we publish is a total or a ratio; not one player name, identity or IP address appears here, and none will.

These numbers will be measured again on 26 October 2026. This refresh rests on a round taken at the same time of day and the same weekday as the one it replaces; a round taken at a different hour reports the clock as growth. The previous figure is not silently dropped — the comparison is written as the difference between the two rounds.