Most neurons in the vertebrate nervous system communicate through action potentials that are typically considered all-or-none events, thus conveying information in a digital manner. However, recent work has suggested the possibility of non-stereotypical, hybrid analog-digital spikes (analog spikes for short) that modulate synaptic transmission, blurring the line between analog and digital signaling. In this study, we consider a model of a binomial synapse, where the probability of vesicle release is determined by a product of two factors: a binary indicator of presynaptic spiking, and a continuous variable (payload) that reflects the shape of the spike, influenced by the value of the presynaptic membrane potential. This framework allows us to model synaptic release with analog spikes and evaluate its effect on synaptic information efficiency. Our analysis focuses on transmitted information per Joule, which we measure as the ratio R := I/E of Shannon mutual information I between the pre- and postsynaptic potential to the energy expenditure E. We optimize both, the spiking threshold and the transfer function that relates stimulus intensity to the spike’s payload, to maximize R. We show that synaptic transmission involving continuous spike payloads significantly enhances information efficiency, particularly under conditions of low noise, compared to purely digital and purely analog information transmission, which are both contained as edge cases in our model. The enhancement arises from a combination of two effects: first, the continuous variability of the spike allows for more nuanced encoding of the presynaptic state compared to stereotypical spiking, and second, a spiking threshold avoids the expenditure of signaling energy below this threshold, thereby reducing energy costs compared to purely analog information transmission. Overall, our model provides a normative account of how the coexistence of analog and digital components in synaptic drive can maximize information efficiency.