As your article explains (but too briefly), epicycles do predict wrong. It predicts correct locations but incorrect phases.
"It was not until Galileo Galilei observed ... the phases of Venus in September 1610 that the heliocentric model began to receive broad support among astronomers."
My understanding is that these had not been observed before Galileo, or at least not observed by many and not long before Galileo's time. In that case, they weren't so much incorrectly predicted, as not observed.
"It was not until Galileo Galilei observed ... the phases of Venus in September 1610 that the heliocentric model began to receive broad support among astronomers."