Extended Data Fig. 1: Pauli errors of single-qubit and two-qubit gates.

Gate errors are benchmarked with simultaneous cross-entropy benchmarking (XEB). Errors of single-qubit gates (blue circles) are obtained by running single-qubit XEB sequences for all 100 qubits simultaneously, while errors of two-qubit gates [red bars, including CZ and CPhase (−0.4)] are averaged over the two-qubit layers used in our experiments. For each two-qubit layer, we run two-qubit XEB sequences simultaneously for all the two-qubit gates in this layer. The maximum number of parallel two-qubit gates in our experiments is 50. The inset shows the cumulative distribution of gate errors, with the dashed lines indicating the median values.