I like the idea of basing seeds on moderate examples of a type of spell; levels 5 to 7 seem ideal.  So using 
interplanar telepathic bond, greater clairaudience etc. are fine.  And I'm not just saying that because of 
auroch's strength! 
One theoretical justification is to say that more advanced techniques (available at higher levels) can replicate and far surpass the primitive techniques used in low-level spells.  Kinda like calculus can do things easily that are very difficult (but possible) with geometry.  Epic seeds use these more advanced techniques; they are all about simplifying, streamlining, and scaling up, and so they use the most robust and efficient base spells that are available.
I'm tempted to say that you shouldn't consider casting times for low level spells. A 
limited wish can replicate a 5th level spell with a 1 hour casting time as easily as a 5th level spell with a standard action casting time.  This suggests that later insights have repaired whatever flaw makes those early attempts so unwieldy to cast in low valence slots, and so, from the standpoint of an advanced practitioner employing epic magic, they are both standard action spells.
Similar remarks apply to troubling features about time-related effects like duration.  If an early spell in a suite lasts for 1 minute per level and a later one lasts for 1 hour per level, then the seed should use a more advanced version.  Or maybe split the difference with 10 minutes per level.  The days to years jump between 
raise dead and 
resurrection can be hand-waved in this fashion.  Transvalent magic literally transcends the limitations of valent magic, especially the limitations of the lower valences.
There's no hard and fast criteria about when you can waive these limitations, but I think that looking at what 
limited wish can duplicate is a good start.  If 
limited wish disregards a feature of a spell (be it gp costs, xp costs, casting time or whatever) then you should at least consider doing the same.  Note that the 
250 300 xp cost is 
instead of a lesser cost, not in addition to.  If only one factor needs to be disregarded and it is a lower level spell, then I'd say do it- the long casting time for 
awaken, say.  If it is at the limit of 
limited wish's power and there are several considerations involved (like for 
raise dead) then you have to be more careful.  I think the current take on the 
life seed is an excellent example of a judicious application of these principles.  
I think, however, that the long casting time of 
divination, contact other plane and 
sending are not, however, particularly relevant to the cost of the seed.