In the development of many AAA popular game titles, the artists went to work first in creating concept art, sometimes in illustrated scene by scene fashion. If it is done very well, then some of those artworks can be used for attracting more investors and for advertising the game once it is done. The more experienced the game development organization, then generally the more feasible it becomes to have a choice in whether to commit to some art assets first or to begin coding. Both could begin literally on the same day, as well. A highly skilled single person Indy developer could likely start with a 3D model or start with assembling some game source code libraries with a bit of coding.
If the first 3D model is exported from the 3D software in one of the standard model file formats according what the game engine will use, then this likely is no problem. One could begin to create a game that is very art asset driven from the start. To get a "map" or "level" rendered by the game engine would be an early priority so all art and coding can build in relation to it.
However, if already existing code libraries are to be used, then it could be preferable to tie these libraries together as a game framework upon which art assets would later be imported. Next, further coding would be needed to make game functionality and gameplay features. It is not unusual for placeholder art to be used until all fundamental game source code is ready.
Some game development companies have been blasted by gamers for using almost the same game engine, game functionality, and similar game concept in consecutive releases from the company. Here is a case where a company feels desperate for improved development cost to sales ratio. In this case perhaps all the basic coding is done and used as placeholder while the company puts new art assets in to the framework and will later add coding as needed.
In my opinion, most beginners to game development should start with coding first and add art assets later. An exception might be the highly skilled 3D artists who needs to build on a level that he or she has made.
My conclusion is that what is made first (art or coding) depends on factors such as personal preference, skill set, available existing framework, and business pressures.