The external validity of dictator games conducted in a lab is often questioned due to the use of small stake sizes that do not correspond to real-world settings. A potential solution to this problem is based on how participant perceptions of stake sizes are affected by their numerical representation. In this paper, I vary the stake size and its numerical representation to examine whether the illusion of large stakes can be created convincingly by implementing inflated numbers through an experimental currency. The share allocated to the recipient does not differ across treatments in this large-sample online experiment. This finding demonstrates that neither an increase in stake size nor a change in its numerical representation influence the share allocated to the recipient in a dictator game.