The Verification Venue · a setting on every phone, answered honestly

Black Costs Less, But Only Just

Dark mode does save battery on an OLED phone, but at the brightness you actually keep your screen, it's a rounding error. The belief isn't a myth; it's mis-scoped. Drag the controls and watch it collapse.

The story everyone repeats: black pixels are "off" on OLED, so a dark theme sips less power. That's true. What nobody says is how much, and when. Purdue engineers wired up four OLED phones (Pixel 2, Moto Z3, Pixel 4, Pixel 5), built a per-frame power model, and measured it. The answer has three moving parts (screen type, brightness, and how much of the screen is dark), so here they are, as three controls. Everything below the instrument is recomputed from its numbers.

Estimated phone power

Dark mode saves

43%

of total phone power, right now

Display is

52%

of total draw at this brightness

Total draw (light)

1.66 W

illustrative: see the check

Auto-brightness parks most people around 30-50% indoors. Drag down and watch dark mode shrink toward nothing.

A dark theme turns most of the screen near-black; a light page is mostly white. This is the "dark mode" knob.

Lever 1: colour (dark mode)

saves 43%

flip light → dark at this brightness

Lever 2: brightness (dim to 50%)

−47%

total power, keeping the current theme

At 100% brightness on OLED, dark mode is a real win: the model lands near 43%, inside Purdue's measured 39-47%. Now drag the brightness down to the 30-50% most people actually use. The saving falls into the single digits: Purdue measured 3-9%. The display's slice of the pie shrank, so turning its pixels off buys you less. The fear was real; it was just aimed at the wrong brightness.

The complete answer is three twists deep

Twist 1: the belief is partly right

Turn the brightness to 100% and dark mode genuinely saves ~40% of your phone's power. The myth isn't fake; it's mis-scoped. It quotes the max-brightness number for the everyday-brightness case. At the 30-50% where auto-brightness actually sits, the same effect is 3-9%: true, but you'd never feel it.

Twist 2: the answer flips on LCD

Toggle the screen type above to LCD and dark-mode savings collapse to essentially zero. An LCD has a single backlight behind every pixel; it stays on at full power whether the pixel shows white or black; the liquid crystal just blocks the light it doesn't emit. Black pixels cost the same as white ones. "Does dark mode save battery" has no single answer until you know which screen you're holding.

Twist 3: the real lever is brightness, not colour

Because OLED power climbs steeply with the brightness setting, dimming from 100% to 50% cuts display power about 10×, and it works on any screen, OLED or LCD, in light mode or dark. Watch Lever 2 above dwarf Lever 1 everywhere except max brightness. If you want your battery to last, the slider that matters is the bright one.

Every case at once

Dark-mode savings (light→dark) as a share of total phone power, recomputed live from the model; and, in the last column, what dimming 100%→50% does to display power instead.

Screen · brightnessdark-mode savesdisplay sharedim 100→50%: display

The dark-mode column is small everywhere on LCD and small at low brightness on OLED; the dimming column is huge everywhere. That contrast is the answer.

The check: recomputed in front of you, matched to the paper

The instrument is the Purdue OLED power model, smoothed to two knobs. Display power is P = D₁·g(b)·content(f) with g(b)=b^γ and γ=log₂10=3.3219, so dimming 100%→50% multiplies display power by g(0.5)=0.5^γ=0.1000: exactly the paper's ≈10×. The rest-of-phone baseline is calibrated once, so full-brightness dark-mode savings equal 43%, the midpoint of Purdue's 39-47%. Everything else falls out of that single calibration:

…

Named free choices: light-mode content is taken as 15% dark (mostly white), dark-mode as 90% dark; "dark" pixels keep a 5% residual glow (real dark UIs aren't pure black). The illustrative device is a 1 W full-white display over a 0.80 W baseline: the split, not the watts, is what's calibrated to the paper. This is an estimate, not a meter of your phone; magnitudes vary by device, which is exactly why the paper reports ranges. Run it offline: node research/does-dark-mode-save-battery/verify-does-dark-mode-save-battery.mjs.

What's idealised here, and what's solid

Solid. On OLED, each pixel emits its own light, so display power really does scale with how much of the screen is lit; turning pixels black really does save power. Brightness really is the dominant lever, and it climbs faster than linearly with the setting. On LCD the backlight is a single always-on lamp, so pixel colour barely moves the meter. These three facts are the load-bearing physics and they're not in dispute.

Idealised / estimated. Purdue's real model (PFOP) is per-pixel and nonlinear across the RGB colour space; we smooth it to a single "fraction dark" knob and a power-law brightness curve tuned to reproduce their two headline numbers (≈10× for dimming, 43% at full brightness). The exact percentage on your phone depends on its panel, its brightness-to-luminance curve, the app, and how "black" your dark theme actually is: hence ranges, not a point estimate. "LCD ≈ 0" means backlight-dominated, not literally zero: pixel electronics draw a hair less, so the model shows a few tenths of a percent, not a hard 0.

Out of scope. This is strictly about battery power. Dark mode's effects on eye strain, sleep and readability are separate, contested questions we make no claim about here.