I've negotiated codeshare agreements with carriers of all sizes and at the end of the day, I've found that not only does nobody understand all the possible loopholes and methods to execute these agreements, but I usually pick up a new idea at the end of the day as well.

That said, a majority of codeshares fall under 3 broad formats - a) Hard block, b) Soft block and c) Free sale.

A "hard block" is when Airline A (marketing carrier) buys a block of "x" seats from Airline B (operating carrier) and re-sells them under Airline A's flight number. Airline A pays Airline B for the fixed number of seats every day regardless of whether or not they are filled.

A "soft block" comes in different variants, but the most common type is that Airline A guarantees Airline B a certain number of seats sold over a certain period and is liable for that value at minimum. The difference between this and a hard block is that a soft block allows the variable demand curve to be more optimally accomodated.

"Free sale" is the most common form of modern codeshares, but requires a fairly complex synching of reservation systems. Essentially a free sale entitles both parties to sell "last seat availability" in all fare classes on a flight and for Airline A to pay Airline B at a prenegotiated SPA (Special Prorate Agreement) for each segment flown in a specific fare class. Alternatively, Airline A and Airline B may have a negotiated "override commission" percentage whereby Airline A retains anything from 7% to 20% of the value of the negotiated fare as their commission for marketing the segment.

There are plenty of variations on the above as well, with combinations and different aspects of each mixed together, as well as more complex setups like the opaque/transparent "reverse codeshare wetlease" arrangements. However, the above 3 are pretty much the basis upon which 90% of the world's codeshare agreements are based in some form or the other.