A simple example is if you were playing competitive rock-paper-scissors. The GTO strategy is 33.3% rock, 33.3% paper, 33.3% scissors. If you do anything else as your strategy, you open yourself up to getting exploited. You can extend this to more complicated games like poker, which has a much more complicated GTO strategy.
The other NL modules (B and Tourney) use 2.5x rfi in most situations. The exceptions are in some short stacked cases (min-raise), or when opening from the small blind (3x, 3.5x, or 4x). The 3-bet sizes are also usually 3x when IP (in position) and larger when OOP (out of position). The exceptions are when stack sizes are shorter. In those cases, a smaller 3bet size is chosen to always leave raise-fold as a reasonable option for the solver.
When A and B versions of a module exist, they use different sizings for raises/3bets/4bets/etc.
The PLO modules use pot-sized raises. PLO HU A uses a raise/fold strategy from the SB. PLO HU B uses a limp/raise/fold strategy from the SB.
In RNG mode, each question comes with a randomly generated number from 0-100. The higher the number, the more you should lean towards the more aggressive action.
For example: suppose the optimal strategy is 30% Fold, 25% Call, 45% Raise. In the default mode, the correct answer is Raise since that has the highest %. In RNG mode, if the RNG is < 30, then the correct answer is Fold. If the RNG is between 30 and 55 the correct answer is Call. If the RNG is > 55 (30+25), then the correct answer is Raise. This mode helps you to implement the mixed strategies with the right frequencies.
For 6-max PLO: the PLO (micro) rake levels are 4.5% with a 10bb cap (~$10 PLO). The PLO (mid) rake levels are 5% with a 2.5bb cap (~$100 PLO). The PLO (high) rake levels are 5% with a 0.6bb cap (~$500 PLO).
For HU PLO: the PLO (mid) rake levels are 5% with a 0.625 cap (~$200 PLO). The PLO (high) rake levels are 5% with a 0.3bb cap (~$500 PLO).
Most NL cash game modules are no-rake modules, but we've also added NL 100 (mid), which has rake of 5% up to a 2.5bb cap (~$100 NL).
In simulations with rake, the solver naturally assumes all other players are also rake-aware (folding when they should, cold-calling less, etc). For example, in a raked NL game the solver will open wider in earlier positions and prefer opening hands like A5o (better for blocking our opponent's 3-bet range) over hands like 65s (better for postflop play/board coverage).
If players in your game are not adjusting accordingly (i.e. too much cold calling with not enough 3-betting or 4-betting), then using no rake ranges (rfi/3bet/etc.) in practice would work better. In general, we would recommend starting with no-rake ranges to have a good baseline/fundamental strategy and then explore the differences in raked simulations to see what deviations are made when there is rake.
We (in green) opened to 2.5x and villain (in orange) 3bet to 6x.
Now our options are to fold (and lose the 2.5x we've put into the hand), call the raise of 6, or go all-in for 25.
A more complicated example:
Call 37.5 implies that the villain's 4bet size is to 37.5x.
Fold 12 implies that our previous raise was to 12x.
So we can infer that the hand began with villian opening. Then we 3bet to 12x. Now villain 4bets to 37.5x and the action is on us. We can either fold (and lose the 12 we've put into the pot already), call the 37.5, or go all-in for 100.
At the moment you can assume all scenarios involve at most two players (with everyone else folding). We plan to add more multiway scenarios in the future.