Every guide on how to increase FPS in CS2 opens with a block of console commands to paste. The problem is that several of the most commonly recommended ones do not exist in CS2 at all. The game accepts nothing, changes nothing, and says nothing, so you conclude the machine is the problem.

Ghost commands still being recommended

We checked these against the game’s own command and variable dumps rather than against other articles. The first four are leftovers from the previous game, still being republished as current advice.

That matters beyond the wasted minute. Pasting a list of dead lines and seeing no improvement is how people talk themselves into buying hardware they do not need.

The four settings that actually move frames

Almost all of the achievable gain lives in the video options rather than in the console.

Resolution. The single biggest lever. Every pixel costs work, and dropping a step down is worth more than every console line combined.

Shadow quality. The most expensive individual setting relative to what it adds. Also the one worth thinking about twice, since shadows carry information about players you cannot see.

Effect and particle detail. Cheap to lower and mostly invisible during a fight.

Multisampling. Edge smoothing costs a great deal on weaker hardware and changes nothing about whether you hit anything.

Change these one at a time, not all at once. Otherwise you have no idea which of them did the work and you end up playing on unnecessarily ugly settings.

What you can realistically gain

Worth an honest number, because guides promising to double your frame rate are selling something.

On a machine that already runs the game acceptably, going through the video settings sensibly buys somewhere in the region of twenty to forty percent. Noticeable, and not transformative.

The bigger gains almost always come from somewhere else: a background application eating the processor, an overlay hooked into the game, or a power profile throttling the machine.

Check those before touching graphics settings. A browser with thirty tabs and a chat client with hardware acceleration on can cost more than every video option combined.

Settings that cost more than they give

A short list of things people disable and then wonder why they keep dying.

Shadows at the lowest setting remove a source of information. Seeing a shadow around a corner is a genuine advantage and you are trading it for a handful of frames.

Dropping the resolution too far makes distant players smaller and harder to distinguish from the background. There is a point where the extra frames do not compensate for not seeing the target.

Turning off the game’s own frame counter is fine. Turning off every visual cue you use to read the game is not, and that is the trap in guides written as if only the number mattered.

Commands worth having in the config

Two exist and both are useful, which is a short list compared with what the average guide offers.

The frame counter tells you what any change actually did, and without it you are guessing based on feel, which is unreliable in exactly the situations you care about.

The frame cap is the more interesting of the two. Set slightly above your monitor’s refresh rate, it stops the card rendering frames nobody will ever see, which lowers heat and, on laptops, keeps the machine from throttling mid match.

Both belong in a file that reloads at every launch, so an update cannot quietly reset them. How to build one is in our guide to the CS2 autoexec.

How to measure whether anything changed

1. Turn on the counter and note the number in a fixed spot on a map you know.
2. Change one setting. One, not five.
3. Return to the same spot and compare.
4. Test through smoke and in a firefight, since that is where drops actually hurt.
5. Keep whatever helped and put the rest back.

An empty server makes this considerably easier, because nothing else is moving and the numbers are comparable between runs. The setup takes two commands and is covered in our walkthrough of the CS2 practice server config.

The number that matters is not the average

Two machines showing the same average frame rate can feel completely different to play on, and this is why.

What breaks aim is not a low average, it is a sudden drop at the moment something happens: a smoke lands, four players enter your view, a flash goes off. The average absorbs those moments and hides them.

So the useful measurement is the worst moment in a round rather than the number sitting on screen while you stand in spawn. If the counter reads two hundred and dips to sixty when a smoke deploys, that dip is what you are actually playing on.

Which changes what is worth adjusting. Effects and particles matter more than they appear to, because they are precisely what is being drawn during those dips.

Launch options, and how few of them matter

The second half of most FPS guides is a string of launch options, and that list has shrunk far more than the articles admit.

The options that genuinely do something now are unglamorous: forcing a refresh rate the game did not detect, or pointing the game at your config file. Neither of those adds frames, they simply stop the game from behaving unexpectedly.

The rest of the usual string, the memory flags and the threading switches, comes from an engine that no longer runs this game. Steam accepts anything you type, and that is why those lines survive: nothing ever tells you they were ignored.

A short launch line is easier to debug for the same reason a short config is. When something breaks after an update, you want a field you can read at a glance rather than forty characters you copied in 2019.

Frames and how the mouse feels

An unstable frame rate rarely gets blamed for what it actually causes, because the symptom shows up somewhere else entirely.

When the picture stutters, the same hand movement lands differently on screen, and that reads as a mouse problem rather than a performance one. People then start changing sensitivity, which makes an already inconsistent situation worse.

The tell is where it happens. Sensitivity behaves identically everywhere on the map. Frame drops cluster in specific places and specific moments, usually the busy ones.

Fix the picture first and only then touch the mouse, using the measurement in our guide to CS2 sensitivity. Tuning aim on an unstable frame rate means tuning against a moving target.

Where settings stop and hardware starts

The part nobody can fix with a config, and it is worth recognising it rather than chasing settings forever.

CS2 leans on the processor harder than the previous game, so on older machines that is the bottleneck rather than the graphics card. The symptom is characteristic: lowering graphics settings changes almost nothing, because the constraint is elsewhere.

You can recognise it easily. If your frame rate is roughly the same on lowest and on medium settings, going lower buys nothing and costs you image quality for free.

In that situation the only thing that genuinely helps, short of new hardware, is taking load off the processor: closing what runs in the background and disabling overlays. Everything else is decoration.

Monitor and refresh rate

Frames your screen cannot display do not exist as far as your eyes are concerned, and they still cost the card work.

If the screen refreshes sixty times a second and the game renders three hundred frames, you see sixty. The rest turns into heat. That is the argument for a cap set a little above the refresh rate rather than at maximum.

The exception applies to players who feel the difference in input latency at very high frame rates. That effect is real and applies to a minority, so treat it as something to test rather than as a rule to follow.

Frequently asked questions about increasing FPS in CS2

Which commands actually increase FPS?

Essentially none. The frame counter and the frame cap exist and are useful. The classic list of optimisation commands mostly does not exist in CS2.

Does lowering resolution help most?

Yes, it is the single largest lever, and also the one that most affects how easily you can see distant players.

Should I turn shadows off completely?

Only if you need the frames badly. Shadows carry information about players you cannot see directly.

What should I set the frame cap to?

Slightly above your monitor’s refresh rate. Rendering far beyond it produces heat rather than visible frames.

Why did nothing change after I pasted a config?

Most likely because the commands in it do not exist in CS2 and were silently ignored.

Are drops in smoke normal?

Common, yes. Effects and particle settings are the ones that affect those moments most directly.

How do I know if my processor is the bottleneck?

If lowest and medium settings produce roughly the same frame rate, the graphics settings are not what is limiting you.

Last updated: September 2026

The list of commands that do not exist comes from comparing common guides against the game’s own command and variable dumps, checked in September 2026. Valve changes both between builds, so verify in the console before keeping any line.

CS2 sensitivity changes your game more than every other setting combined, and it is the one setting you cannot usefully copy from anybody else. A different mouse, a different mousepad and a different hand produce a different result from identical numbers.

Three numbers people keep confusing

DPI belongs to the mouse and describes how many signals it sends per inch of movement. You change it in the mouse software or with a button on the shell.

In game sensitivity is the multiplier the game applies to that signal. It defaults to 1.25 and the permitted range runs from effectively zero up to eight.

eDPI is the two multiplied together. Eight hundred DPI at 1.0 sensitivity is identical to four hundred DPI at 2.0 as far as the game is concerned.

Which is why comparing raw sensitivity values is meaningless. Somebody playing at 0.8 can be turning faster than somebody at 2.0 if their mouse is set four times higher.

The measurement that beats any config

Rather than hunting for the correct number, measure what your hand is actually doing.

1. Lay a ruler along your mousepad and set the mouse at the start of it.
2. Turn your character exactly one full rotation without lifting the mouse.
3. Measure the distance travelled. That is your centimetres per 360.
4. Write it down along with the DPI and sensitivity that produced it.
5. Compare it against the range most competitive players occupy, roughly thirty to fifty centimetres.

That figure is the only one that transfers between people, because it describes arm movement rather than equipment. If yours comes in under twenty centimetres you are aiming with your wrist and losing accuracy at range, which is the single most common problem among players who cannot climb.

Picking a starting point

Choose one row and stay in it for two weeks. Jumping between values every evening is the most efficient way to learn none of them.

CS2 sensitivity while scoped

A separate setting, and the one people trip over most, because its name suggests something other than what it does. When you scope in, the game applies its own multiplier. It defaults to one, with a range from 0.01 to 3.

A value of one does not mean the movement on screen is identical to hip fire. The field of view narrows, so the same hand movement sweeps a different angle, which is where the feeling that the mouse suddenly accelerates comes from. The practical approach: leave it at one until it genuinely bothers you. If you consistently overshoot with a sniper rifle, drop toward 0.8 and test it against bots rather than in a match.

Now the part no config includes.

The setting nobody writes about

The game carries a separate scale for the vertical axis, defaulting to one, with a range from zero to two. At one, up and down are as sensitive as left and right. Below one the vertical axis slows down, which some players find helps with recoil control, since a spray pulls the crosshair up while you pull it down.

It is not a shortcut to accuracy and it is not for everybody. It is worth knowing it exists, because plenty of people fight their recoil without realising the axes can be separated at all.

A value of zero locks vertical movement entirely, so if somebody’s mouse has mysteriously stopped looking up or down, this line pasted from a config is usually the culprit.

Two more lines to delete.

Two commands that do nothing

Both still circulate widely and both are dead.

The raw input command does not exist in CS2. It is a menu option now, and the config line does nothing at all.

A separate scoped multiplier ending in “mouse” does not exist either, despite appearing in guides. Only the base one works.

We checked both against the game’s own variable dump. It is the same mechanism that leaves other people’s configs full of lines from the previous game, covered in more detail in our guide to the CS2 autoexec.

Windows breaks more than the game does

Before hunting for a perfect number, check that the operating system is not rewriting the signal on the way through.

The pointer speed slider has eleven positions. Only the sixth, the middle one, passes the signal through untouched. Every other position rescales movement and destroys repeatability.

Second, enhanced pointer precision, which is acceleration. With it on, the same hand movement produces a different turn depending on how quickly you make it, which is the exact opposite of what you want.

Third, mouse software with its own acceleration or smoothing. Open it once and turn off anything that sounds like assistance.

Frame rate changes how sensitivity feels

Before concluding your mouse is too slow or too fast, rule out the thing that impersonates a sensitivity problem.

With an unstable frame rate, the same hand movement produces different results on screen because the image cannot keep up. It reads as a mouse that drags in one moment and skates in the next, so people start changing sensitivity instead of looking for the cause.

You can tell them apart easily. Sensitivity behaves the same everywhere. Frame drops happen in specific places on the map or through smoke.

What to do about it, and which settings genuinely add frames, is in our guide to increasing FPS in CS2. Stabilise the picture first, then tune the mouse.

Grip is the other half of the equation

The same centimetres per turn produces very different results depending on what moves the mouse.

Wrist aiming gives you a few centimetres of travel and excellent precision on small corrections. Arm aiming gives you a dozen or more, slower but far steadier across long sweeps.

Most people use both alternately, which is the crux: a sensitivity tuned for the wrist kills arm movement and the reverse is equally true. Extreme values only work for players who genuinely commit to one.

If you are unsure which you are, look at your mousepad after an hour. Wear concentrated in a small patch means wrist. Wear spread across half the surface means arm.

Changing mouse without losing your aim

New hardware is where most people quietly ruin their settings, because they carry over the numbers instead of the movement.

A different mouse at a different DPI with the same in game value gives a different turn speed. Rather than copying the figures, measure centimetres per turn again and adjust the sensitivity until you land on the number you had before.

A change of mousepad works the same way, more subtly. The hand travels differently across a different surface, and after a few days most people want a few centimetres more or less than they did.

Which is the argument for keeping your figure written down alongside the settings in your config file. It is the one value you will not reconstruct from memory.

Testing whether a change helped

Any new sensitivity feels good at first, because novelty forces concentration. That wears off in a few days and only then does the truth appear.

A reasonable test is twenty minutes of deathmatch and one question afterwards: were you overshooting the target more often, or undershooting it. Overshooting means too fast, undershooting means too slow.

The second test concerns turning around. If you need two mouse lifts to face behind you, you have gone too far the other way and will pay for it every time somebody flanks.

Both want an empty server with bots, since in a match you are also measuring somebody else’s play. The commands for that are in our walkthrough of the CS2 practice server config.

Frequently asked questions about CS2 sensitivity

What DPI should I use?

Usually 400 or 800. The product of DPI and sensitivity is what matters, so DPI alone decides nothing beyond how finely the mouse measures movement.

What eDPI should I aim for?

Most players sit between 400 and 1600. Rather than targeting a number, measure your centimetres per full turn.

How many centimetres per 360 is right?

Typically thirty to fifty. Below twenty you are aiming with your wrist, above sixty you lose the ability to turn behind you.

Should I change the scoped multiplier?

The default of one is a fine starting point. Change it only if you consistently overshoot while scoped.

Why does my mouse feel different after a Windows change?

Because outside the middle slider position the system rescales the signal. Set the sixth of eleven positions and disable enhanced pointer precision.

Can vertical sensitivity be set separately?

Yes, the game has a vertical axis scale ranging from zero to two. Zero locks vertical movement completely.

How long does adapting to a new sensitivity take?

Around two weeks of regular play. Changing it again before that means starting from zero.

Last updated: September 2026

Defaults and ranges come from the game’s own variable dump, checked in September 2026. The distances quoted per style are observations from published player settings rather than a rule, so treat them as a starting point for your own measurement.

Dust 2 callouts are usually presented as one enormous image with sixty labels on it. That is a reference, not a guide. In a real match perhaps twenty of those names get said, and the difference between knowing them and knowing all sixty is nothing at all.

The twenty that carry the map

Learn that column and you can hold a conversation with any team on this map. Everything else in the sixty name image is either a subdivision of one of these or a name that fell out of use two versions ago.

A side, where most rounds start

Long is the corridor down the right hand side from the attacking spawn, and the doors at its entrance are the single most contested object on the map. Half of all rounds are decided by who takes them and when.

Pit is the sunken area on the left as you come out of Long, and the reason a player holding it can survive an entire push without being seen from the site.

Car sits at the entrance to the site itself. Ramp is the raised approach on the far side, usually held after the bomb is planted. Site covers the plant area, and nobody bothers subdividing it during a fight.

Short, and the name nobody agrees on

Short is the zigzag from the middle up to the A site, and it is the second route into the same place, which makes controlling both simultaneously the defining problem of defending A.

Xbox is the crate near the middle that you jump on to reach Short. The name is older than most players and it has never been translated, adapted or replaced.

Goose is the passage under A coming from the middle, named after a developer of the series. It is also the area three separate groups of players call three different things, so if a call confuses your team, this is probably the one.

Ninja is the alcove between Short and the site where a defender waits without making noise, and the name tells you everything about its purpose.

Middle, the part that decides rotations

Mid is the open channel between the two halves of the map, and control of it decides whether defenders can reach the far site in time.

Top mid is the upper end of it on the attacking side. Double doors is the gap through to the defending side, usually shortened to doors. Catwalk is the raised walkway leading toward Short.

Lower tunnels connects the middle to B and is the reason the middle matters to both sites at once. A call from here is normally worth more than a call from either bomb site, because it arrives before anything has been committed.

B side and the tunnels

Fewer names, more improvisation, because there are simply fewer distinct places.

The tunnels are the main route to B from the attacking side and split into upper and lower. That distinction is the whole point: it tells the defence which direction an entry is coming from and how long they have.

Back plat is the raised platform at the rear of the site. Window is the opening on the defending side. Big box and the stack of crates on the site are where the defence sits, and in most teams all of it collapses into “by the boxes” regardless of the official names.

Dust 2 callouts that include a timestamp

Here is the part missing from every guide with a map image, and it matters more than any name on this page.

A place is half the information. The other half is time: when you saw them, whether they are still there, whether they have moved. “One Long” on its own means almost nothing ten seconds later, because by then he is somewhere else entirely and your teammate is walking into a spot you described in the past tense.

The complete version is short and contains three elements: how many, where, when. “Two Long, just now.” “One under A, a while ago.”

Add one word about status: dead, or simply out of sight. That is the line between information and a guess, and a team plays the round differently depending on which one it received.

How to learn them in twenty minutes

Load the map alone and walk the attacking routes, naming each place out loud as you pass it. Saying it aloud attaches the word to the image far faster than reading a diagram, however silly it feels in an empty server.

Then the same run through the middle, and again through the tunnels to B.

Then walk both routes from the defending side. The same places viewed from the opposite direction look different and get confused differently, so one pass does not cover it.

After that, go into a normal match and talk, even hesitantly. A name someone corrects mid round sticks permanently. A name read off an image is gone by the evening.

An empty server for that takes two commands and no downloads, and the setup is in our walkthrough of the CS2 practice server config.

Common mistakes on voice

Sentences that are too long. By the time you finish, the situation has changed. Three words is the target.

Calling from spectator without saying so. You are looking through someone else’s eyes, so say “from spec”, or your team assumes it is your own current information.

Directions instead of places. “On the right” means something different to every player depending on where they stand.

Silence after a kill. Where you killed someone and what they had is as useful as any warning, and it costs one short sentence.

Calling as a defender is a different job

Attacking calls describe where you are going. Defending calls describe what is arriving, and that changes what is worth saying.

The valuable information is a count and a direction early enough for someone to rotate, which usually means calling before the fight rather than after losing it. “Three tunnels” said while they are still in the tunnels is worth a round. The same words said after you die are a status update.

The other half is what you did not see. Silence from the far side of the map means nobody has checked it, and saying so explicitly stops your team from assuming it is covered.

What this changes in ranked

You can play entire seasons without saying a word, and plenty do, so it is worth naming what that costs.

Without communication you are playing five separate matches at once. Everyone sees a fragment of the map, nobody assembles it, and the team reacts late to everything.

Callouts are the cheapest fix available, because they require vocabulary rather than aim or reflexes. That is why they sit alongside map knowledge in our breakdown of what actually moves your rating, in CS2 ranks explained.

Once this map is comfortable, the usual next one is at the other end of the pool, and we have collected those names separately in Mirage callouts.

Names that have fallen out of use

Older guides, particularly forum threads that have been online for over a decade, carry names describing objects the map no longer has.

Using them makes you sound authoritative to nobody and confuses everybody under thirty. The map has been rebuilt more than once and the vocabulary moved with it, while the threads stayed exactly where they were.

The rule of thumb: if you hear a name once in ten matches, you do not need it. Twenty names in constant use cover the entire map, and the remainder is trivia for people who enjoy trivia.

Frequently asked questions about Dust 2 callouts

How many callouts do I actually need?

About twenty. Long, doors, pit, car, Short, xbox, goose, mid, catwalk, double doors, upper and lower tunnels, window, back plat and the two sites cover nearly everything.

What is goose on Dust 2?

The passage under the A site from the middle, named after a developer of the series. It is also the name most likely to be misunderstood.

What is the difference between upper and lower tunnels?

Lower comes from the middle, upper from the attacking side. For the defence that difference tells them which entry to expect.

Why is a crate called xbox?

Because of its shape, and because the name has been in use since long before CS2. Nobody has ever renamed it.

Should I learn callouts from an image?

An image is a starting point. The names stick when you walk the map and say them out loud.

What do I say when I do not know the name?

Describe it in one phrase with a reference point, such as “by the boxes on B”. Better than silence every time.

Have the callouts changed since CS:GO?

Almost none of the current ones. The layout survived, so old names still work, though some have quietly stopped being used.

Last updated: September 2026

Names collected from English and Polish map references and from what is actually said on servers, checked in September 2026. Several places carry more than one accepted name and none of them is the single correct one.

A CS2 autoexec is a plain text file the game reads at launch. Building one takes a minute. The interesting problem is that a large share of the configs circulating online contain commands CS2 no longer has, and the game says nothing when it skips them.

Where the file goes

One path, no exceptions. Right click CS2 in your Steam library, choose Manage, then Browse local files. From the folder that opens, go into game, then csgo, then cfg. That is the destination.

Create a text file there and name it autoexec.cfg. When saving from Notepad, set the file type to All Files, because otherwise Windows appends .txt and the game will never see the file.

The fastest way to check: turn on file extensions in Explorer. If what you see is autoexec.cfg.txt, you have already found the reason nothing loads.

Making the game read it

The file alone is not enough, and this is where most people stop without realising it. The game does not scan the folder hoping to find configs, so you have to point at it directly.

Open the properties for CS2 in Steam, find the launch options, and enter +exec autoexec.cfg.

The extension matters. The version without .cfg appears in older guides and in CS2 it frequently does nothing at all.

At the end of the file itself, add host_writeconfig. That forces your current settings into the game’s own config, which keeps them from evaporating on exit.

Proving the CS2 autoexec loaded

Do not infer it from your settings. Make the game say so out loud, which takes one line and saves half an hour of guesswork.

Near the end of the file, before the write line, add something like echo AUTOEXEC LOADED.

Launch the game, open the console, and look for that text. If it is there, the file loaded and any remaining problem is in its contents rather than its location. If it is missing, go back to the filename and the launch option.

How to open the console, and what to do when the key does nothing, is covered separately in our piece on the CS2 console not opening.

One detail that saves restarts: after editing the file, type exec autoexec in the console and it reloads without leaving the game.

Dead commands still doing the rounds

We checked these against the game’s own command and variable dumps rather than against other blogs. All six appear in configs marketed as current, and all six are silently ignored.

That is the argument for writing your own file instead of pasting somebody else’s. A config you did not write is a list of lines you cannot vouch for.

What belongs in a short file

Fifteen lines that change something, rather than two hundred that mostly do not. A short file also has the practical advantage that when something misbehaves, you can read the whole thing at a glance.

Sensitivity and the scoped multiplier, so a change of mouse is one line rather than an evening.

Viewmodel. Field of view and offsets, since this is the only thing permanently covering part of your screen that you control.

Crosshair, if you prefer keeping it in the file rather than in the menu.

A frame cap, set slightly above your monitor’s refresh rate.

Binds for anything you do often that otherwise costs a menu.

Binds are the highest value lines

Display settings are set once and forgotten. Binds fire every round, which makes them the best return per line in the whole file.

The standard set: a jump bound to the scroll wheel, a viewmodel toggle, quick buys, dropping the bomb. None of it is an exploit. Each does exactly what your fingers would do, only faster and without fumbling.

The scroll wheel jump is worth its own explanation, including why it beats the spacebar, and that is in our guide to the CS2 bhop bind on scroll.

One rule when adding binds: one key, one action. Multi state toggle scripts look clever and end with you pressing something other than what you intended in the one round where it mattered.

Pro configs and why they rarely help

Every so often a file appears with a player’s name attached and gets treated as a recipe.

Half of what is inside depends on hardware and habit: a sensitivity tuned to one mouse and one mousepad, a viewmodel set for a particular screen, binds laid out for a keyboard you do not own. The other half is audio and video tuned to a machine unlike yours.

Transplanted whole, that produces an unfamiliar game rather than a better one. The sensible use is to read it and steal individual ideas, usually a bind you had not thought of.

When the file is not the problem

There is a specific case where the echo line appears, the file clearly loaded, and one setting still refuses to work.

Most often the command is restricted on the server. A chunk of the console is limited to your own server or requires server commands enabled, and an official match will refuse it regardless of where you typed it. Which commands those are is set out in our list of sv_cheats 1 commands.

The second cause is a typo. The game does not announce it on screen, it writes a quiet line into the console that scrolled past during map load.

The third, and the most irritating, is two contradictory lines in one file. The lower one wins, so a value set twice gives you the second one while you search everywhere except the obvious place.

A second file for practice commands

Splitting the settings into two files is worth the extra minute, because the things you want while playing and the things you want while practising work against each other.

The autoexec holds what should always apply: sensitivity, viewmodel, binds, frame cap. Practice commands are a different animal. Infinite ammo, grenade trajectories, instant respawn and bot control only function on your own server, and an official match refuses them anyway.

Keeping those in a separate file, called on demand, is cleaner than one long file with half its lines inert most of the time. It also means you can hand that file to a friend without exporting your personal sensitivity along with it.

The full set of commands worth putting in it is in our walkthrough of the CS2 practice server config.

Load order, or why settings revert

The game reads its own config first, then your autoexec if the launch option points at it. That is why the file overrides the menu rather than the other way round.

On exit, however, the current state is written back. Change something in the menu mid session and that value lands in the game’s config, where it will lose to your autoexec at the next launch. It looks random and it is simply ordering.

Keep each setting in one place. Either the file or the menu. Mixing both is the source of every “it changed by itself” report.

One copy, two machines

If you play on more than one computer, the file becomes worth more than the settings inside it.

Keep it in any cloud synced folder and copy it into place after installing the game. Rebuilding every bind from memory takes an evening. Copying a file takes a second.

One value does not travel: sensitivity tuned to one mouse behaves differently on another and needs recalculating rather than copying. Everything else transfers unchanged and gives you your own game on a borrowed machine in about a minute.

Frequently asked questions about the CS2 autoexec

Where do I save autoexec.cfg?

In the game folder, under game, then csgo, then cfg. Browse local files in Steam takes you there fastest.

Does it work without a launch option?

Sometimes, and it is not worth relying on. Adding +exec autoexec.cfg settles it permanently.

How do I confirm the file loaded?

Put an echo line in it and look for that text in the console after launch.

Do I need to restart after editing?

No. Type exec autoexec in the console and it reloads.

What is host_writeconfig for?

It forces current settings into the game’s own config so they persist after you close the game.

Can a custom config get me banned?

Not when it uses commands the game exposes normally. Binds and settings are a standard part of it.

Why do commands from a downloaded config do nothing?

Because some of them are leftovers from the previous game. Check in the console whether CS2 recognises a command before keeping it.

Last updated: September 2026

Paths and commands checked in September 2026 against the current build. The list of commands that no longer exist comes from comparing common configs with the game’s own command and variable dumps, and can change with a major update.