IBM reported a 99.00% mean network-benchmarking fidelity for relay-based quantum-state transfer between two Nighthawk R1 processors. Announced on Oct. 27, the four-link experiment measured a specific transfer operation, not the accuracy of an entire quantum computer.
What the 99.00% measures
The paper reports the mean fidelity across the four L-couplers for relay-based state transfer. Fidelity here is an operation-specific benchmark: the figure does not give the probability that a complete quantum computation will succeed.
The researchers also measured distant CZ gates, a type of two-qubit operation between processor modules. Those gates have a separate fidelity result, so the two measurements should be read independently.
| Operation | Mean fidelity | Best link, including reported uncertainty |
| Relay-based state transfer | 99.00% | 99.29% ± 0.07% |
| Distant CZ gates | 94.75% | 96.42% ± 0.28% |
Two Nighthawk R1 processors, four L-couplers
The experiment linked two superconducting IBM Nighthawk R1 processors, each with 120 qubits, using four L-couplers. A qubit is a basic unit of quantum information. Each coupler joined processor lattice qubits through two frequency-tunable L-qubits, two RF-SQUID-based i2m couplers, and 80 centimeters of superconducting aluminum coaxial cable.
IBM says the chips were full Nighthawk processors, but only a subset of their qubits was wired out for this development experiment. The 120-qubit capacity of each processor therefore does not describe how many qubits participated in the inter-chip connections.
Beyond state transfer
The team also performed distant CZ gates between qubits on separate modules. IBM says the researchers used two adjacent L-couplers at once to generate independent entangled states across the processors. The gate results in the table are distinct from the higher state-transfer result.
A step toward modular quantum computing
Levante-EMV reported that all four links continued working after three heating-and-cooling cycles without rebuilding their connections. IBM describes improving coupler yield and reproducibility, reducing error rates, and scaling to more couplers as future development work. The reported experiment demonstrates inter-processor operations; IBM’s fault-tolerant systems remain a future goal.
