Nvidia Reflex is the single easiest latency win on the green team, but it only does its job if the rest of your system isn’t quietly adding milliseconds behind its back. The priority order for low latency is: enable Reflex (or Reflex + Boost) first, cap your FPS slightly below your monitor’s refresh rate so the render queue stays empty, then eliminate the CPU-side bottlenecks that Reflex can’t fix. A GPU-bound frame with Reflex beats an uncapped frame with a backed-up queue every time, because input lag is mostly a function of how long a frame waits in line, not how fast the GPU can draw it once it starts.
This guide assumes the Nvidia App (or the legacy GeForce Experience per-game settings) on Windows 11, current Game Ready drivers, and a G-SYNC or high-refresh monitor. Everything applies equally to a 2026 driver stack on RTX 30-series through RTX 50-series cards, and most of it applies to GTX 900-series and newer with driver-level Reflex support.
As an Amazon Associate we earn from qualifying purchases at no extra cost to you.
Our Top Picks
Top 3 picks at a glance
Quick Recommended Settings
| Setting | Recommended value | Why |
|---|---|---|
| Nvidia Reflex Low Latency (in-game) | On + Boost (competitive) / On (everything else) | Flushes the render queue just before the CPU submits a frame; Boost holds GPU clocks high to avoid clock ramp delay |
| In-game frame rate cap | 3-4 FPS below refresh rate (e.g., 237 on a 240 Hz panel) | Keeps you GPU-bound, not queue-bound, and prevents Reflex from being overridden by a full queue |
| Nvidia App Max Frame Rate (global) | Off — set per-game only | A global cap silently throttles every title including ones where you want uncapped frames |
| Low Latency Mode (Nvidia Control Panel) | Off if Reflex is available in-game; Ultra only as fallback | Reflex supersedes it; running both wastes frames without reducing latency further |
| Vertical sync | Off in-game; use G-SYNC + cap instead | V-Sync can add a full refresh interval of latency; G-SYNC with a cap keeps tearing away for far less |
| Power management mode | Prefer maximum performance (per-game profile) | Prevents GPU clock drops between rounds that Reflex Boost would otherwise have to paper over |
| G-SYNC | On for single-player; Off for pure competitive titles | Eliminates tearing without V-Sync latency, but any syncing adds a small amount vs. raw uncapped play |
| Windows Hardware-Accelerated GPU Scheduling | On (default) | Reduces driver-side queueing overhead on modern cards |
Setting-by-Setting
Nvidia Reflex Low Latency: On vs. On + Boost
Exact value: Set to On in any title that offers it; use On + Boost for competitive shooters. The toggle lives in each game’s own video or gameplay menu — in Counter-Strike 2 it’s Settings → Video → “NVIDIA Reflex Low Latency”; in Fortnite it’s Settings → Video → “NVIDIA Reflex Low Latency”; in Valorant it’s Settings → Video → General → “NVIDIA Reflex Settings.”
Trade-off: On drains the CPU-side render queue when the GPU is backed up, which is where most avoidable latency lives. On + Boost additionally commands the GPU to hold its boost clocks even during light loads, so the first frame of an engagement isn’t eaten by clock ramp-up (typically a few milliseconds of idle-to-boost transition). Boost costs extra power and heat — irrelevant on a desktop, noticeable on a laptop’s battery and thermals. If the game you play has no Reflex option, this setting does not exist for you and no driver toggle replicates it exactly; see the Low Latency Mode fallback below.
In-game frame rate cap
Exact value: Cap at refresh rate minus 3-4 FPS. On a 240 Hz monitor, set the in-game limiter to 236 FPS; on 144 Hz, use 140 FPS. Use the game’s built-in limiter first (Valorant: Settings → Video → “Max FPS”; CS2: fps_max command, e.g., fps_max 236). If the game lacks one, use the Nvidia App’s per-game “Max Frame Rate” control (Graphics tab → scroll to “Max Frame Rate”) — this driver-level limiter is the lowest-overhead external option and beats RTSS for Nvidia-only setups.
Trade-off: Uncapped FPS feels faster but leaves the GPU saturated, which forces the render queue to fill and hands latency back to you even with Reflex on. Capping 3-4 FPS below refresh keeps the GPU under 100% utilization so Reflex’s queue flush actually has room to work, and leaves headroom for G-SYNC to operate without V-Sync. You lose a handful of average FPS; you gain consistent frame pacing.
Low Latency Mode (Nvidia Control Panel fallback)
Exact value: Off when the game has native Reflex. If the game has no Reflex support, set it to Ultra. Path: right-click desktop → Nvidia Control Panel → Manage 3D Settings → “Low Latency Mode” — set globally, or better, per-game under the Program Settings tab.
Trade-off: Ultra caps the render queue at one frame, approximating what Reflex does. It is not Reflex: it can’t synchronize with the game engine’s frame submission the way the SDK does, so it helps less. Running it on top of Reflex does nothing measurable and occasionally interferes with frame pacing, which is why the recommendation is Off whenever Reflex exists in the title.
Vertical sync and G-SYNC configuration
Exact value: V-Sync Off in every game menu. If you have a G-SYNC monitor: enable G-SYNC in Nvidia Control Panel (Set up G-SYNC → “Enable G-SYNC, G-SYNC Compatible”), keep “Enable settings for the selected display model” checked, and enable V-Sync only in the Nvidia Control Panel (Manage 3D Settings → Vertical sync → On) while leaving it off in-game. This is the standard G-SYNC + cap arrangement: the driver-level V-Sync acts as a backstop at the cap boundary and doesn’t engage during normal G-SYNC operation.
Trade-off: In-game V-Sync adds up to a full refresh interval of latency. The G-SYNC + cap setup removes tearing for roughly one millisecond or so over raw uncapped play. For tournament-style competitive play, some players disable G-SYNC entirely, run uncapped or capped-at-refresh, and accept tearing for the absolute minimum pipeline.
Power management mode
Exact value: Nvidia Control Panel → Manage 3D Settings → (Program Settings tab, select your game) → Power management mode → “Prefer maximum performance.”
Trade-off: GPUs downclock aggressively during low-load moments (menus, quiet rounds, spectating). The first frames after a burst start on low clocks. “Prefer maximum performance” pins clocks high; combined with Reflex + Boost it’s belt-and-suspenders, but alone it fixes clock-ramp latency in games that lack Boost support. Cost: higher idle-in-game power draw and temperatures. Set it per-game, not globally — a global setting keeps your card at high clocks on the desktop for no benefit.
Hardware-Accelerated GPU Scheduling
Exact value: On. Windows Settings → System → Display → Graphics → “Change default graphics settings” → Hardware-Accelerated GPU Scheduling → On, then reboot. On current Windows 11 builds this is on by default; verify rather than assume.
Trade-off: HAGS lets the GPU manage its own VRAM scheduling, trimming driver overhead. Occasionally a specific game or capture tool misbehaves with it; if a title stutters with HAGS on, test off, but leave it on system-wide otherwise.
By Hardware Tier
Low-end (GTX 16-series, RTX 20-series, 1080p/60-144 Hz)
Your frame rate is the bottleneck, so protect every millisecond: Reflex On (Boost only if thermals allow), cap at refresh minus 3, Low Latency Mode Off, textures and shadows lowered first since they’re VRAM-bandwidth heavy. On a 60 Hz panel especially, the cap-3-below rule matters less than keeping the GPU under 97% utilization — drop one demanding setting (usually shadows or ambient occlusion) until GPU usage in a fight sits in the 80-90% range. Don’t run Ultra Low Latency Mode and expect Reflex-like results; it’s a partial fix.
Mid-range (RTX 30/40-series, 1440p/144-240 Hz)
You have headroom, so the priority is consistency: Reflex + Boost, cap at refresh minus 3-4 via the in-game limiter, G-SYNC on with driver V-Sync on, HAGS on. If you use DLSS frame generation in single-player games, note that Reflex is mandatory with frame gen — the driver enables it automatically — and frame generation itself adds latency, so never use it for competitive titles regardless of how good the FPS number looks.
High-end (RTX 40/50-series, 240 Hz+ or 4K high-refresh)
At very high frame rates the render queue is rarely the problem — clock stability and display pipeline become dominant. Reflex + Boost on, Power management “Prefer maximum performance” per-game, cap at refresh minus 4 to keep G-SYNC in range, and consider disabling G-SYNC entirely for competitive play since tearing at 300+ FPS is barely perceptible while syncing overhead is not. Check that your monitor is actually running its maximum refresh: Windows Settings → System → Display → Advanced display → “Select a refresh rate” — a surprising number of high-refresh panels ship at 60 Hz out of the box.
Common Mistakes
Stacking Low Latency Mode Ultra on top of Reflex. What causes it: guides written before Reflex existed still recommend Ultra as the universal fix. How to check: open the game’s video menu — if “NVIDIA Reflex Low Latency” appears and is set to On or On + Boost, the driver-level Low Latency Mode should be Off. What to do: Nvidia Control Panel → Manage 3D Settings → Program Settings → select the game → Low Latency Mode → Off. How to undo: set it back to Ultra (or Use global) if you ever disable Reflex in a title without it.
Capping FPS at exactly the refresh rate with G-SYNC on. What causes it: capping at the panel’s maximum pushes frames into the V-Sync backstop path, adding latency spikes. How to check: compare your in-game FPS counter (Alt+R for the GeForce overlay’s statistics, or Ctrl+R in the Nvidia App overlay) to your refresh rate; if they touch, the cap is too high. What to do: lower the cap by 3-4 FPS. How to undo: raise the cap back if you later disable G-SYNC.
Leaving a global Max Frame Rate limit on. What causes it: setting a driver-level cap globally for one game and forgetting it. How to check: Nvidia App → Graphics → Global settings → look at Max Frame Rate. What to do: set it to Off globally; apply caps per-game only. How to undo: re-add the cap to the specific game profile.
Playing on a monitor running below its native refresh. What causes it: fresh Windows installs and DisplayPort cable swaps resetting to 60 Hz. How to check: Windows Settings → System → Display → Advanced display → “Select a refresh rate.” What to do: set it to the panel’s maximum; if the option is missing, verify the cable is DisplayPort-certified for the bandwidth and reselect the mode in the Nvidia Control Panel → Change Resolution. How to undo: change it back if you swap monitors.
Enabling frame generation in competitive games. What causes it: the FPS counter looks spectacular. How to check: in the game’s upscaling menu, look for “Frame Generation” — if it’s on in a shooter, that’s the problem. What to do: turn it off; keep DLSS Super Resolution on Quality or Balanced if you need the performance. How to undo: re-enable it for single-player titles where the added latency doesn’t matter.
FAQ
Does Reflex work on GTX cards?
Partially. Reflex requires an Nvidia GPU from the GTX 900 series onward with current Game Ready drivers, and the game must implement the Reflex SDK. On GTX cards you get the queue-flushing “On” behavior; “Boost” clock control is reserved for RTX-series hardware. Check the game’s own video menu — if the option is grayed out or absent, your card or driver doesn’t support it in that title.
Should I use Reflex + Boost always?
Use it whenever your GPU has thermal and power headroom — virtually all desktops. Skip Boost on laptops running on battery or in hot environments, where pinning clocks costs runtime and adds fan noise without changing the outcome much. Boost provides no benefit when your GPU is already near maximum utilization, because clocks are already pinned by load.
Does Reflex help if I’m already CPU-bound at high FPS?
Yes, and this is a common misunderstanding. CPU-bound is precisely the condition where the GPU finishes frames faster than the CPU submits them — meaning the queue is short and Reflex has little to flush, but also meaning your latency is already low. Reflex matters most in GPU-bound scenarios: graphically heavy scenes, explosions, smoke. Enable it regardless; it costs nothing and protects the worst-case moments.
Does Reflex lower my FPS?
Reflex On can reduce average FPS slightly in GPU-bound scenes — usually a small percentage — because it deliberately throttles CPU submission to keep the queue empty. That trade is the entire point: you exchange a sliver of throughput for consistently lower input lag. Boost costs FPS only in the sense that it holds clocks (and power draw) high during light loads; in heavy scenes it’s effectively free.
