BeamNG.drive on a Mac
MacBook Pro, M3 Pro
BeamNG is not a racing game but a deformation simulator: a car is solved as a soft body of hundreds of nodes, and the processor does that work, not the graphics card. So the story here is different from a shooter, and the number means something different too.
The numbers
A MacBook Pro with an M3 Pro, medium-high settings, ordinary driving around a map in a Hirochi.
| Scene | Frames |
|---|---|
| A Hirochi, medium-high | 100–110 fps |
That is ordinary play, not a benchmark loop. Heavy maps hit the graphics card exactly as they would on a modest PC: BeamNG can load any hardware if you ask it to.
Why it is an easier game here than a shooter
BeamNG has no kernel-level anti-cheat — and that is what half the Windows games on a Mac stumble over. Nothing stands in the way of running it, and there is no account to put at risk.
BeamNG's interface is a built-in browser, the same engine launchers are made of. It works here: panels and menus draw properly, while the game itself is drawn by its own engine.
What you need
- A Mac with Apple Silicon — M1 or newer. Intel Macs are not supported.
- macOS 14 or later.
- Your own Steam account with BeamNG.drive on it.
- Disk space: here the game takes 54 GB — that is what Steam itself reports and what Finder measures. Keep some room to spare: updates ask for more on top.
How it goes
- Sign in on this site. Your profile shows a download link and an activation code.
- Install Uncork, enter the code, and it builds a bottle with Windows Steam inside.
- Sign in to Steam yourself and install the game as usual. Your Steam password and Steam Guard code are typed only into Steam's own window — Uncork never asks for them.
- Launch. The first run takes a while as shaders build.
Two of you on one Mac
The game itself has a multiseat mode: several players, each with their own controller and their own vehicle. Turn it on in Options → Gameplay → "Use multiseat for this session". The camera stays single and pulls back to keep both cars in frame, so you cannot drive far apart.
Real split screen comes from SplitNG in the official repository: up to four players, each with their own camera and instruments. Start free roam, connect the gamepads and press Start on the second one — the screen splits. A keyboard and a gamepad also count as two players, so you can try it alone.
It works under Uncork — tested on an M3 Pro with two PlayStation 5 controllers over Bluetooth. The mod hooks into an extension point inside the game and carries no native code, so there is little in it to break.
It costs frames, and the price is honest. Split screen draws the scene twice, with every shadow and reflection pass, plus a second vehicle with its own physics.
Two players on Italy — one of the heaviest maps in the game — got 19–27 FPS with traffic on and 20–40 once it was off, on settings lowered below the usual. For comparison, a single player on Johnson Valley, a desert and a far simpler scene, got about 40. Different maps, so those numbers are not a before-and-after of split screen but two separate situations.
So before playing together, turn traffic off first: every traffic car is drawn in every pass of the scene and simulated on top of that — and the two of you are each other's traffic anyway. Then dynamic reflections and detailed mirrors: those are whole extra passes over the scene, and split screen doubles them along with everything else.
Sparkles on the foliage
White dots appear on bushes and grass: they shimmer as the camera moves and are worst in bright sun. This is not an artifact of translating DirectX to Metal — the same thing shows up on Windows, just less often.
Two settings in the game itself fix it, both under graphics:
- Turn HDR off. The extended range stretches highlight brightness, and a dot that was nearly invisible becomes a white flash.
- Anti-aliasing SMAA → TAA. SMAA reads a single frame and mistakes one bright dot for an edge. TAA looks across several frames in a row and puts that dot out — precisely because it flickers.
Checked on an M3 Pro: after those two switches the sparkles are gone completely. Frames do not drop for it — TAA is cheaper than SMAA, and HDR off removes a tone-mapping pass on top.
Mods: telling a good one from plasticine
The game has a built-in mod store, and that is where everyone goes first. The catch is that real brands are not there: BeamNG buys no licences, so anything called “BMW” or “Porsche” has in fact been exported from another game, Automation. Those cars have smooth, sparse geometry — hence the plasticine feel.
The numbers show it. We counted triangles in the game's own cars: the median is 209 thousand, the simplest is the Moonhawk at 110 thousand, the most detailed the D-Series at 775 thousand. The Automation export we measured came to 51 thousand — twice as crude as the worst stock car.
Hand-built mods live not in the store but as attachments in forum threads on beamng.com. The same numbers tell the difference: a Countach from there is 2.2 million triangles, a BMW E34 1.6 million, an Audi R8 1.4 million — three times the most detailed stock car.
How to tell before you install
A mod archive is an ordinary zip, and you can look inside before installing:
- Files named
camso_mark an Automation export. Of the 18 mods we took from the built-in store, all 18 had them. - A
vehicles/folder has to be there. If the archive holds further archives, the game will not unpack them and no car appears at all. Two “packs” caught us that way: listed among the mods, adding nothing. - Interior parts in the jbeam — the words
interior,dashboard,seat. Without them the shell is empty: there is a body, but nowhere to sit.
A broken dashboard costs more than it looks
A mod car's instrument screen is not a picture but an embedded Chromium: a real page that Lua feeds thirty times a second. When the page cannot be found, the game quietly substitutes a placeholder — but Lua does not know that and keeps calling functions on it. Every call throws.
In one session of ours that came to 1,075 errors from the Audi R8's navigation and 926 from the dashboard that five mods borrow from the stock ETK800. One Chromium process died outright, and the game's log hit its 15,000-entry ceiling and began truncating itself — so it stopped being useful for diagnosing anything else.
The causes were dull: one mod's path pointed at a file absent from its own archive, another at a folder under a different name, a third at a neighbouring car's page, which is only mounted while that car is in the session. None of this is Mac-specific — the same breaks on Windows, where people simply look at the log less often.
The cure is one line in the mod's jbeam: the htmlFilePath field. If you
like the car but its dashboard is black, that is almost certainly where it went.
Measured on an M3 Pro overnight on 21 September 2026: 50 mods downloaded and taken apart, triangles counted in every model and the file list read. The figures above come from those measurements, not from general reasoning.
What we have not measured
- Pile-ups with ten cars at once. Soft-body physics becomes processor-bound there, and we have not checked what is left of the frame rate.
- Large maps such as Italy and Utah at maximum settings.
- Other Macs. An M1 Air will not behave like an M3 Pro, and we have not tested one.
- Mods. There are thousands, and we have run none of them.
If it comes out differently on your machine, write and say so — measurements that only hold in one place are worth correcting.