B Grade B
IBM Heron r3: QUOPS
204 QUOPS score
As stated in the source: “Q = 204 at 3.1x10^5 QUOPS/s (width 6)”
Caveats
Preprint (v1, 10 Sep 2026), not peer reviewed. New benchmark defined and run by a team that includes the maker of two of the four machines (Quantinuum). Scores depend on compilation/routing effort chosen by each team. Rate methods differ by platform (reserved-window wall clock for Quantinuum; dedicated rate run with 500k shots for Willow; QPU usage time with batch approximation for ibm_boston), so rates are not strictly like-for-like. Run dates not given; rate uses an approximation across interleaved shapes.
- Set-up
- ibm_boston. Physical qubits, width w=6, shape (6,204); 100 circuits per shape and 200 reference circuits per width; qiskit transpiler (sabre layout/routing, optimization level 3), dynamical decoupling added before execution; Heron r3 processor. Rate 3.1x10^5 QUOPS/s from QPU usage time with batch approximation.
- Method
- QUOPS v1 (arXiv:2609.12146v1), physical-qubit architecture
- Compared only with
- results in the group “v1 physical”
- Measured
- Not stated
- Published
- 10 September 2026
- Measured by
- Paper authors (Sandia National Laboratories-led team incl. Quantinuum and NVIDIA co-authors), using IBM Quantum resources of the Air Force Research Laboratory
- Independent of the maker
- Yes
- Peer-reviewed
- No
- Repeated by others
- Not yet
Source
Benchmarking the computational power of quantum computers, arXiv. Preprint. DOI: 10.48550/arXiv.2609.12146
Comparable results
| Machine | Value | Set-up | Published | Measured by | Grade | Record |
|---|---|---|---|---|---|---|
| Quantinuum Helios | 1,504 | v1 physical | 10 Sep 2026 | Maker | B | r-000048 |
| Quantinuum H2 | 1,320 | v1 physical | 10 Sep 2026 | Maker | B | r-000050 |
| Google Willow | 216 | v1 physical | 10 Sep 2026 | Maker | B | r-000051 |
History
- 27 October 2026: added
Spotted a mistake? Report an error, quoting r-000052.