I personally feel that any kind of payment per episode model will fail. Who wants to pay upwards of $25 for one season of a show? If you rent those episodes, you can never go back and watch them again for free.
Hulu and Netflix grant you all-you-can-eat access for one flat rate. This is what the cable companies have done and have been successful too.
Until Apple creates a subscription plan for ALL tv shows, count me out.
Apples and oranges. There are several emerging models, and some people will favor one over the other, and most people, I suspect, will choose a mixture.
Model 1: Pay to own. Download it. Keep a copy of it forever. If you want to watch that one episode of Hannah Montana 10 years from now, you can do it for free! (minus the aggregate cost of having to have maintained the video and the risk of the video being unplayable at that point).
Model 2: Subscription fee for partial library, streamable. Library is limited, and different subscription services overlap significantly. I suspect this will either move towards more focused subscriptions (all ABC shows for $5/month) or a single consolidated service (Hulu wins all and offers every show under the sun for $20/month). My money's on more focused subscriptions, although Hulu would obviously want you to believe they'll have everything under their umbrella. The ad-support versus fee-support of this model is variable right now, but conventional wisdom has the mix here spreading more towards the Cable content/ad mix (two minutes of show for one minute of ad watching), perhaps in return for more content without fees rising.
Model 3: Subscription fee for full library, high quality. This is the Cable / Satellite model. $60-100 per month for a large library, pushing $200/month for "everything". Plus one minute of ads for every 2 minutes of show. You can time-shift high quality, and you can stream stuff you forgot to time shift. Obviously Model 2 and Model 3 are in direct competition.
Model 4: Pay to own, high quality. This is the DVD/BluRay model. Similar to 2/3, this is in the most direct competition with Model 1. The advantage of (1) over (4) is, however, immediacy of availability (I can't imagine a studio putting out a DVD each week the night the show airs, can you?)
There are probably several other models which will come out, but the main orthogonal aspects I'm seeing right now are:
1. Quality. Streaming versus downloadable versus physical delivery.
2. Immediacy. Moment of airing versus next day versus next week versus next Fall.
3. Cost model. Convenience of a set monthly bill versus the flexibility of pay-for-what-you-use. I think, in general, the pendulum is swinging away from all-you-can-eat for everything (cell phones, internet, you name it); things like this where your usage is highly visible and controllable are the most likely places for all-you-can-eat models to falter (unlike internet, say).
4. Breadth of content / granularity of payment. Do you pay for a single episode, a single show/season, everything a particular "network" puts out, or some larger collection? People have been asking for a la carte cable subscriptions for years and years; seems odd that we have the technology to deliver that model finally now and no one would want it any more. This is obviously fairly related to (3).
5. Ownership. When you stop paying, does everything go away, or do you just stop receiving new content?