Why to use plot.ly over bokeh?



We’re developing a data visualization tool in Python to show temporal series from user activity in certain websites.
We’re currently studying which technology better suits our needs, before to start to code.

We need the stats to be interactive, the ability to filter showing data by time (start date - end date) and to mix graphs generated from different datasets into a single chart. Preferable, it would run in a web server. The webapp would consist in a dynamic number of tabs (no browser tabs, but webapp’s own tabs), each one showing a particular subset of charts with different metrics being shown.
The dataset can be quite big (hundreds of MB in csv format), but several graphs should be generated in a tolerable amount of time for the user (<1min).

With this requirements, I’m considering bokeh vs plot.ly overall. Why should I use plot.ly over bokeh? which advantages does it give to me?