One thing I hate about people who are strongly anti-torture is the statement: "torture doesn't work, people will say anything to stop the pain".
"People will say anything to stop the pain"? Very true, and "anything" includes the truth.
If you have two subjects: Torture them separately until their stories match exactly, start with little weird details that are unconnected to what you want to know (so any prepared story is useless). Any inconsistency results in more pain for both until stories match, continue as needed and slowly work your way to the info you really want. By the time you are asking questions you care about they are too afraid of what will happen if they lie for them to even think about trying. Despicable, but works like a charm. For only one subject you need some other way to confirm what they say, this means that speculative torturing for information does not work well or at all for single subjects, you need at least two.
In order to torture anyone effectively, you need to know what the "correct" answer you're trying to get is ahead of time. Even in your (laughably convenient) hypothetical situation, wherin you have two 100% certainly enemy agents who each have 100% accurate knowledge about things 100% relevant to your goal, whom have had not any opportunity to collaborate beforehand, it is more than plausible that the leading questions you will ask them will give them enough information for them to guess what you want to know.
The Nazis were not known for being reserved. Nor were they notable for great espionage. However, one element of military intelligence they excelled at: interrogating enemy pilots. How?
By giving them booze and nature walks.
Torture is not effective, and the myth that it is effective is built on buying one's own propoganda -- that the enemy is cowardly and weak, when in fact they are likely brave (albeit misguided) to face the United States in war and strong enough to survive a battle against her finest. Torture is furthermore not only ineffective (in that it mainly allows one to "confirm" a hunch), but also insidious. It contaminates intelligence organizations that use it, because one or two lucky breaks where a hunch proved correct after being "supported" by torture make it seem credible. Thereafter, other resources are put under pressure to remain accurate while being as fast and cheap as torture. More and more, the agency guesses and tortures, then makes new guesses from that torture-confirmed "intel." The witchhunts are self-perpetuating.
I could copy-paste the wikipedia page on torture, but I'll let you research for yourself. Suffice it to say that torture won the Japanese Empire such war-winning intel as the fact that the USA had 100 additional nuclear bombs and that Tokyo and Kyoto were next (at the time, the USA had expended the two bombs it had and the soldier being tortured had no idea what an atom bomb even was).
But let's put all that aside. Some will say that, if we assume torture is 100% effective, then it should be used if it would save lives. This is a sound argument, but it is not valid, as torture is not 100% effective. In fact, its use makes people less safe, as intelligence agencies agree.
In short, someone arguing for torture as a means of gathering intelligence is very much like someone advocating rape as a contraceptive; a perfect trifecta of ineffectiveness, logical invalidity, and breathtaking moral failure.