Caffeine is an element that is consumed worldwide. It is present in many products such as beverages, chocolate, coffee, tea, energy drinks and medicines. Portable 3D devices working together with colorimetric and fluorimetric reactions have been able to determine the presence of caffeine in different kinds of samples. Also, digital image-based methods using smartphones have conferred portability and accessibility to miniaturized devices that are innovative and promising options for quick and low cost analyses. This study proposes a miniaturized fluorimetric device to determine caffeine by digital image using a smartphone. The OpenCamera app was used to capture images that were processed using ImageJ software to obtain RGB channels values. The red (R) channel signal intensity was selected as the analytical response. The device developed was applied to determine caffeine in an energy drink and medicines. The method developed presented a linear range from 100 to 600 mg L of caffeine, and quantification (LOQ) and detection (LOD) limits of 100 mg L and 30.0 mg L, respectively. The caffeine concentration found in the products analyzed was 328 mg L (±2.5%) for the energy drink, 345 mg L (±15%) for medicine A and 322 mg L- (±7.3%) for medicine B. The proposed device presented important characteristics such as low cost, required small volumes of reagents and samples, quick analysis, portability and suitable to be applied in complex matrices.