There is a long thread on timetriallingforum about VO2max training that Andrew Coggan is involved with.
This post by him seems to me to be a reasonable take on VO2 workouts
"Science says that longer intervals are better. (Despite what TrainerRoad might have you believe, 1 min intervals are not a VO2max intervals, unless you combine such shorter efforts with short rest periods.)
My experience says that 10 x 3 min on, 1.5 min off produces different results (in terms of performance ability) than 6 x 5 min on, 2.5 min off. After realizing that, I never did the former workout ever again.
Coincidentally, I was thinking about this thread and in fact this exact question a day or so ago, and concluded that 4-6 min intervals are optimal…3 min is too short, and 8 min (Seiler’s favorite) is too long.
As for the intensity, as hard as you can go while still (mostly) completing the workout “in style” (i.e., w/o struggling mightily or having to reduce the power significantly during the last effort). For most, that will be somewhere between 106% and 120% of FTP, but for some it will be higher, and possibly in a few, slightly lower."
Also worth noting this anecdote from Coggan
"ETA: Re. point #1: once upon a time, I reasoned that since I could maintain ~90% of VO2max for >1 h, I should focus on raising my “ceiling”. 18 wk of 3 d/wk Hickson intervals later, I hit my highest VO2max ever (5.45 L/min at 67.5-68 kg). I then raced the Texas state TT, and got my butt kicked.
Moral of the story? Train for performance, and let your physiology sort itself out. In particular, don’t underestimate the importance of specificity (specificity, specificity, specificity, specificity…is there an echo in here?)."