We introduce another variant of Quantum MIP, where the provers do not share entanglement, the communication between the verifier and the provers is quantum, but the provers are unlimited in the classical communication between them. At first, this model may seem very weak, as provers who exchange information seem to be equivalent in power to a simple prover. This in fact is not the case—we show that any language in NEXP can be recognized in this model efficiently, with just two provers and two rounds of communication, with a constant completenesssoundness gap. Similar ideas and techniques may help help with other models of Quantum MIP, including the question of noncommunicating provers with unlimited entanglement.