So first let me try to clear the confusion.
A synthesizer or sampler is multitimbral if it is capable of producing more than one type of sound or timbre (pronounced tam bur) at a time. Usually this is described as the number of “parts” a unit can play at once. For example, a Kurzweil K2500 is 16-part multitimbral, meaning it can produce 16 different sounds at once (a sound being defined as a single patch or preset; part one might be piano, part two strings, part three trombone, part four flute, and so on. Generally these parts are assigned to different MIDI channels for independent control).
Emphasis on the "generally", but this does not itself mean multi-timbral.
Tempest is a 32-part multitimbral 6-voice synth. In one "beat" it can have have 32 different "sounds" (patches) loaded up (16 beats loaded at the same time). So if you are controllings "sounds" via MIDI then you have access to 32 different patches, all on one channel. So everything is on ONE channel, EXCEPT! one sound. You can set up one sound on a different channel, allowing it to be played as a polyphonic synth over MIDI.
This is just the MIDI setup of the Tempest. The builtin sequencer however can play every sound at any key.