The standard sample rate for video is 48 kHz, so that's what I use. Generally speaking, the audio is going to end up as a compressed format, so the bit depth of the source audio won't matter too much. Either 16 or 24 bit should be fine, but 32 bit floating point is probably overkill and likely to add more processing load without much if any sonic benefit.
If you're going to mix in stereo, be sure to mix mono compatible. Some people are going to consume your content on phones with one speaker, so if there are phase issues between left and right channels, it might not sound good on a mono speaker. Also, if you want bass instruments (kick drum, bass guitar etc.) to be audible on small speakers, they will need some presence in their upper harmonic ranges. In my opinion, a good mix will do all that by default.
Make sure your levels are healthy but not clipping. It might be worth putting a mastering limiter on your output and gently raising the level of the whole mix to around -14 LUFS while also assuring there's a full 1 dB of True Peak headroom.
