IBM announced the IBM Quantum Nighthawk r3 superconducting processor, targeting 10,000 accurate quantum operations across 520 programmable qubits built from 2,034 physical qubits with 10 l-couplers for chip-to-chip interconnects. Two r3 chips are slated for integration into a 1,040-qubit system by 2027. The currently deployed Nighthawk r2 features 120 programmable qubits, 218 couplers, and 120 reset elements (458 total physical elements) in a square-lattice topology with four-neighbor connectivity, accessible via the IBM Quantum Platform as ibm_phoenix. Nighthawk r2 achieves over 100,000 circuits per second — roughly 25× the throughput of Heron — by replacing conditional reset with a per-qubit dissipative reset gadget using tunable couplers to a cold environment, reducing effective T1 from ~200 µs to ~25 ns and idle time between shots to ~1 µs. IBM reports 34 billion circuits executed on r2 to date. Partner-reported speedups over Heron include E.ON at 15× for quantum machine learning on energy grids, Oak Ridge National Laboratory at 12× for magnetic materials simulation, Algorithmiq at 8×, and Q-CTRL reducing a 60-qubit fermionic simulation runtime from 4 minutes 25 seconds to 23 seconds with sub-1% deviation. IBM demonstrated accurate observable estimation on circuits exceeding 7,500 gates. Chip-to-chip links using 80 cm cables between two Nighthawk r1 chips achieved up to 99.3% state transfer fidelity, averaging 99%, below the 99.9% target. The IBM Quantum Platform now integrates HPC and quantum workloads, enabling workflows such as sample-based quantum diagonalization to run ~5× faster with 100× more computations, and Cleveland Clinic/RIKEN processing 608 million samples 33× faster via automation.