PhoneArena reported on November 6, 2026, that a WekiHome teardown identified the iPhone Duo’s main-camera sensor as a Sony IMX954 in 1/1.59-inch format. The report puts its sensor area roughly 30% below the Sony IMX905 attributed to the iPhone 18 Pro. Those percentages use opposite baselines, so “about 30% smaller” and “about 40% larger” can both describe the same approximate comparison.
For broader context on the phone, see NeoTeo’s earlier iPhone Duo hands-on coverage.
What the sensor report says
PhoneArena attributed the IMX954 identification and 1/1.59-inch format to a WekiHome teardown. It also reported that the iPhone 18 Pro uses a Sony IMX905 main-camera sensor and that the Duo sensor has roughly 30% less area. The November 6 date is when PhoneArena published its account; it is not a date for the teardown.
A separate estimate from videowien puts the Duo sensor near 1/1.56 inch and about 48 mm², compared with an estimated 71.5 mm² for the iPhone 18 Pro. Those figures imply about 32.9% less area for the Duo: 1 − (48 ÷ 71.5). The creator describes the Duo figure as an inference, so this estimate does not establish an exact sensor model or dimension.
Why 30% smaller can mean about 40% larger
Percentages change when the reference point changes. If a smaller sensor were exactly 30% below a larger one, it would be 70% of the larger sensor’s area. Reversing the comparison, the larger sensor would be about 42.9% above the smaller one: (1 ÷ 0.70) − 1.
That hypothetical result is close to PhoneArena’s “about 40% more” phrasing. It is an arithmetic illustration, not a new measurement; PhoneArena’s figures are approximate.
When comparing percentage changes, check which device supplies the baseline before treating two rounded figures as a contradiction.
Sensor area is not a photo-quality score
A larger sensor area alone does not determine the finished photo. PhoneArena author Victor Hristov described the Duo’s photos as not much worse than the iPhone 18 Pro’s, though still not quite the same. That is his qualitative impression, not a quantified measure of how much image quality differs.
