Two machines returning different answers tells you very little if they weren't given the same problem. @QuipNetwork launched QVM on the Aglais testnet on October 20. Developers describe an optimization task as constraints, then the toolchain compiles three components: An encoder prepares the model. A verifier checks whether the answer satisfies the requirements. A decoder turns the result into something an application can use. Today, the system supports D Wave quantum hardware and classical solvers. IonQ support is still planned. What I'd watch next is independent developers running identical problems across different machines and publishing comparable results. One shared problem definition makes those comparisons easier to inspect. It doesn't automatically make quantum hardware faster or better. It gives developers a more consistent way to find out.