Separate frequencies should be able to interoperate if the decoder chips have narrowband filters that properly reject the other frequencies, which means reading the datasheets to find out the frequency response curves.
This will only work if only one transmitter is operating at a time. The frequency in question isn't the frequency of the light, but the frequency of the pulse-train used to encode the signals. Two simultaneous pulse-trains of different frequencies get mixed up in such a way that it is not possible to separate them afterwards.
As an analogy, imagine sending morse-code messages with a piano. You can send two simultaneous messages, as long as you use different pitches for different messages. The recipients can tell which is their message by listening out for tones of the right pitch (perhaps with the help of an audio filter). But if you try to transmit two simultaneous messages by hitting a single key, even if they are sent at different rates, it's impossible to disentangle the notes into intelligible code. It is
possible to use a single key to send messages to two different recipients if the senders agree to transmit their messages one after the other. The recipients can work out which message belongs to them by listening to how fast the code is sent.
In this analogy, the piano pitch is the wavelength of the IR light, and the rate at which the morse-code is sent is the IR pulse frequency.