Let's say I have 5 models that I am drawing (using the same shader program, different uniform values) in 5 draw calls.
What is faster?
A. Have 5 VBOs, one for each of the 5 draw calls.
B. Aggregate the VBOs into one larger buffer, and then 5 times, draw a specific sub-range from that.
In case it matters: models are small, and target is OpenGL-ES3.