[ { "id": "r-000001", "machine_id": "google-willow", "benchmark_id": "logical-error-rate", "method_version": "Surface code memory, logical error per cycle, neural-network decoder (offline)", "comparison_group": "logical-error-rate@surface-code-d7-memory", "value": 0.00143, "value_text": "0.143% ± 0.003% error per cycle (distance 7)", "unit": "error per cycle", "uncertainty": { "type": "other", "low": 0.0014, "high": 0.00146, "note": "0.143% ± 0.003% as stated" }, "configuration": "Distance-7 surface code memory using 101 qubits; QEC cycle time 1.1 µs; offline neural-network decoder. A distance-5 code with a real-time decoder reached 0.35% ± 0.01% per cycle.", "measured_on": null, "published_on": { "date": "2024-12-09", "precision": "day" }, "source": { "url": "https://www.nature.com/articles/s41586-024-08449-y", "doi": "10.1038/s41586-024-08449-y", "title": "Quantum error correction below the surface code threshold", "publisher": "Nature (Springer Nature); preprint arXiv:2408.13687", "type": "peer_reviewed" }, "measured_by": "Google Quantum AI", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "This is one logical qubit stored as a quantum memory, not a logical qubit doing computation. The figure uses an offline decoder; real-time decoding was only shown at distance 5 with a higher error. The authors say rare correlated error events, about one per hour, limit performance.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000001/" }, { "id": "r-000002", "machine_id": "google-willow", "benchmark_id": "logical-error-rate", "method_version": "Surface code memory, logical error per cycle, AlphaQubit2 neural-network decoder, after reinforcement-learning control fine-tuning", "comparison_group": "logical-error-rate@surface-code-d7-memory", "value": 0.000772, "value_text": "7.72(9) × 10^-4 per cycle (distance-7 surface code)", "unit": "error per cycle", "uncertainty": { "type": "other", "low": 0.000763, "high": 0.000781, "note": "Stated in parenthesis notation as 7.72(9)×10^-4; the type of interval is not stated in the part read." }, "configuration": "Distance-7 surface code on Willow, decoded with AlphaQubit2; control parameters fine-tuned by reinforcement learning on top of conventional calibration.", "measured_on": null, "published_on": { "date": "2025-11-11", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2511.08493", "doi": "10.48550/arXiv.2511.08493", "title": "Reinforcement Learning Control of Quantum Error Correction", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Google Quantum AI and Google DeepMind", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Preprint, not peer reviewed (no journal reference listed on arXiv at v4, 19 Jun 2026). Quantum memory only, not logical computation. Result depends on a neural-network decoder and RL tuning.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000002/" }, { "id": "r-000003", "machine_id": "google-willow", "benchmark_id": "logical-error-rate", "method_version": "Color code memory, logical error per cycle, neural-network decoder (AlphaQubit)", "comparison_group": "logical-error-rate@color-code-d5-memory", "value": 0.011, "value_text": "ε5 = 0.0110(2) per cycle (distance-5 color code)", "unit": "error per cycle", "uncertainty": { "type": "other", "low": 0.0108, "high": 0.0112, "note": "Stated as 0.0110(2); the type of interval is not stated in the part read." }, "configuration": "72-qubit Willow-generation chip. Distance-5 color code logical qubit using 19 data qubits and 18 auxiliary qubits (37 physical qubits) on a 72-qubit Willow processor.", "measured_on": null, "published_on": { "date": "2025-05-26", "precision": "day" }, "source": { "url": "https://www.nature.com/articles/s41586-025-09061-4", "doi": "10.1038/s41586-025-09061-4", "title": "Scaling and logic in the color code on a superconducting quantum processor", "publisher": "Nature (Springer Nature); preprint arXiv:2412.14256", "type": "peer_reviewed" }, "measured_by": "Google Quantum AI", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Different, smaller Willow chip (72 qubits) from the 105-qubit Willow on the spec sheet. Quantum memory result; error suppression from distance 3 to 5 was only Λ = 1.56(4), weaker than the surface code result.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000003/" }, { "id": "r-000004", "machine_id": "google-willow", "benchmark_id": "xeb", "method_version": "Linear cross-entropy benchmarking on random circuit sampling", "comparison_group": "xeb@103-qubits-depth-40", "value": 0.001, "value_text": "XEB fidelity = 0.1% at 103 qubits, depth 40", "unit": "fidelity", "uncertainty": null, "configuration": "Random circuits on 103 qubits, depth 40, on Willow Chip 2 (mean simultaneous two-qubit iSWAP-like error 0.14%).", "measured_on": null, "published_on": { "date": "2024-12-09", "precision": "day" }, "source": { "url": "https://quantumai.google/static/site-assets/downloads/willow-spec-sheet.pdf", "doi": null, "title": "Willow Spec Sheet", "publisher": "Google Quantum AI", "type": "vendor_technical" }, "measured_by": "Google Quantum AI", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Vendor spec sheet with no uncertainty given. At this scale the XEB fidelity cannot be checked by full classical simulation; it is an estimate. The spec sheet's claim of 5 minutes on Willow vs 10^25 years classically is a separate estimate not recorded here.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000004/" }, { "id": "r-000005", "machine_id": "google-willow", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Randomized benchmarking, reported as average error", "comparison_group": "two-qubit-gate-fidelity@mean", "value": 99.67, "value_text": "0.33% ± 0.18% two-qubit gate error (CZ), mean, simultaneous", "unit": "%", "uncertainty": { "type": "other", "low": 99.49, "high": 99.85, "note": "converted from the stated error and its uncertainty" }, "configuration": "CZ gates, mean across device, all gates operated simultaneously.", "measured_on": null, "published_on": { "date": "2024-12-09", "precision": "day" }, "source": { "url": "https://quantumai.google/static/site-assets/downloads/willow-spec-sheet.pdf", "doi": null, "title": "Willow Spec Sheet", "publisher": "Google Quantum AI", "type": "vendor_technical" }, "measured_by": "Google Quantum AI", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "This is an error, not a fidelity (equivalent to about 99.67% mean fidelity). Recorded exactly as stated. Vendor-reported; Chip 1 and Chip 2 are different physical chips with different numbers. Converted from the stated mean CZ error of 0.33% ± 0.18%.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000005/" }, { "id": "r-000006", "machine_id": "google-willow", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Randomized benchmarking, reported as average error", "comparison_group": "two-qubit-gate-fidelity@mean", "value": 99.86, "value_text": "0.14% ± 0.052% two-qubit gate error (iSWAP-like), mean, simultaneous", "unit": "%", "uncertainty": { "type": "other", "low": 99.808, "high": 99.912, "note": "converted from the stated error and its uncertainty" }, "configuration": "iSWAP-like gates, mean across device, all gates operated simultaneously.", "measured_on": null, "published_on": { "date": "2024-12-09", "precision": "day" }, "source": { "url": "https://quantumai.google/static/site-assets/downloads/willow-spec-sheet.pdf", "doi": null, "title": "Willow Spec Sheet", "publisher": "Google Quantum AI", "type": "vendor_technical" }, "measured_by": "Google Quantum AI", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "This is an error, not a fidelity. Different gate type (iSWAP-like) and a different chip from the CZ figure on Chip 1, so the two are not directly comparable. Vendor-reported. Converted from the stated mean error of 0.14% ± 0.052%.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000006/" }, { "id": "r-000007", "machine_id": "google-willow", "benchmark_id": "randomized-benchmarking", "method_version": "Single-qubit randomized benchmarking, reported as average error", "comparison_group": "randomized-benchmarking@single-qubit-mean-simultaneous", "value": 0.00035, "value_text": "0.035% ± 0.029% single-qubit gate error, mean, simultaneous", "unit": "error per gate", "uncertainty": { "type": "other", "low": 6e-05, "high": 0.00064, "note": "0.035% ± 0.029% as stated" }, "configuration": "Single-qubit gates, mean across device, simultaneous operation.", "measured_on": null, "published_on": { "date": "2024-12-09", "precision": "day" }, "source": { "url": "https://quantumai.google/static/site-assets/downloads/willow-spec-sheet.pdf", "doi": null, "title": "Willow Spec Sheet", "publisher": "Google Quantum AI", "type": "vendor_technical" }, "measured_by": "Google Quantum AI", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Error, not fidelity. Vendor-reported. Chip 2 lists 0.036% ± 0.013%.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000007/" }, { "id": "r-000008", "machine_id": "google-willow", "benchmark_id": "physical-qubit-count", "method_version": null, "comparison_group": "physical-qubit-count@gate-based", "value": 105, "value_text": "105 qubits", "unit": "qubits", "uncertainty": null, "configuration": "Square-grid layout, average connectivity 3.47 (4-way typical).", "measured_on": null, "published_on": { "date": "2024-12-09", "precision": "day" }, "source": { "url": "https://quantumai.google/static/site-assets/downloads/willow-spec-sheet.pdf", "doi": null, "title": "Willow Spec Sheet", "publisher": "Google Quantum AI", "type": "vendor_technical" }, "measured_by": "Google Quantum AI", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "A count, not a performance measure. Google has also used a 72-qubit Willow processor (color code paper), so 'Willow' does not always mean 105 qubits.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000008/" }, { "id": "r-000010", "machine_id": "ibm-heron-r3", "benchmark_id": "layer-fidelity", "method_version": "Error per layered gate (EPLG) on a 100-qubit chain; EPLG_100 = 4/5(1 - LF^(1/99))", "comparison_group": "layer-fidelity@eplg-100-ibm", "value": 0.002694, "value_text": "2Q error (layered) 2.694E-3", "unit": "error per layered gate", "uncertainty": null, "configuration": "ibm_boston. Average EPLG in a layered chain of 100 qubits, from IBM's current calibration data.", "measured_on": null, "published_on": null, "source": { "url": "https://quantum.cloud.ibm.com/computers?view=table", "doi": null, "title": "IBM Quantum Platform: Compute resources (table view)", "publisher": "IBM Quantum", "type": "vendor_technical" }, "measured_by": "IBM Quantum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Live calibration value that changes over time; the page shows no timestamp. Snapshot read on 27 Oct 2026. Recorded as EPLG (an error); layer fidelity itself is not shown. Lower is better. Live value from IBM's calibration table, read on 27 Oct 2026; IBM updates it regularly.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000010/" }, { "id": "r-000011", "machine_id": "ibm-heron-r3", "benchmark_id": "clops", "method_version": "CLOPS (IBM definition, layers of a 100x100 circuit; page heading 'CLOPS (or CLOPS_h)')", "comparison_group": "clops@ibm-clops-h", "value": 340000, "value_text": "340K", "unit": "layers per second", "uncertainty": null, "configuration": "ibm_boston. As listed on IBM Quantum Platform compute resources table.", "measured_on": null, "published_on": null, "source": { "url": "https://quantum.cloud.ibm.com/computers?view=table", "doi": null, "title": "IBM Quantum Platform: Compute resources (table view)", "publisher": "IBM Quantum", "type": "vendor_technical" }, "measured_by": "IBM Quantum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Live value, rounded to 340K, with no timestamp; snapshot read 27 Oct 2026. IBM's docs do not say whether this is the original CLOPS or the newer CLOPS_h variant. Other Heron systems show 300K–340K. Live value from IBM's calibration table, read on 27 Oct 2026; IBM updates it regularly.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000011/" }, { "id": "r-000012", "machine_id": "ibm-heron-r2", "benchmark_id": "layer-fidelity", "method_version": "Error per layered gate (EPLG) on a 100-qubit chain", "comparison_group": "layer-fidelity@eplg-100-ibm", "value": 0.003439, "value_text": "2Q error (layered) 3.439E-3", "unit": "error per layered gate", "uncertainty": null, "configuration": "ibm_kingston. Average EPLG in a layered chain of 100 qubits, from IBM's current calibration data.", "measured_on": null, "published_on": null, "source": { "url": "https://quantum.cloud.ibm.com/computers?view=table", "doi": null, "title": "IBM Quantum Platform: Compute resources (table view)", "publisher": "IBM Quantum", "type": "vendor_technical" }, "measured_by": "IBM Quantum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Live calibration value with no timestamp; snapshot read 27 Oct 2026. Best-performing Heron r2 in the table at that moment (ibm_fez 4.436E-3, ibm_marrakesh 4.696E-3). Live value from IBM's calibration table, read on 27 Oct 2026; IBM updates it regularly.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000012/" }, { "id": "r-000013", "machine_id": "ibm-nighthawk-r2", "benchmark_id": "layer-fidelity", "method_version": "Error per layered gate (EPLG) on a 100-qubit chain", "comparison_group": "layer-fidelity@eplg-100-ibm", "value": 0.002606, "value_text": "2Q error (layered) 2.606E-3", "unit": "error per layered gate", "uncertainty": null, "configuration": "ibm_phoenix. Average EPLG in a layered chain of 100 qubits, from IBM's current calibration data.", "measured_on": null, "published_on": null, "source": { "url": "https://quantum.cloud.ibm.com/computers?view=table", "doi": null, "title": "IBM Quantum Platform: Compute resources (table view)", "publisher": "IBM Quantum", "type": "vendor_technical" }, "measured_by": "IBM Quantum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Live calibration value with no timestamp; snapshot read 27 Oct 2026 while the system showed 'Paused (maintenance)'. EPLG on a square lattice (Nighthawk) and heavy-hex (Heron) measures a 100-qubit chain on each, so the figure does not capture Nighthawk's extra connectivity. Live value from IBM's calibration table, read on 27 Oct 2026; IBM updates it regularly.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000013/" }, { "id": "r-000014", "machine_id": "ibm-nighthawk-r2", "benchmark_id": "clops", "method_version": "CLOPS (IBM definition; page heading 'CLOPS (or CLOPS_h)')", "comparison_group": "clops@ibm-clops-h", "value": 2000000, "value_text": "2.0M", "unit": "layers per second", "uncertainty": null, "configuration": "ibm_phoenix. As listed on IBM Quantum Platform compute resources table; Nighthawk r2 uses fast dissipative qubit reset (idle time between circuits as low as 1 µs).", "measured_on": null, "published_on": null, "source": { "url": "https://quantum.cloud.ibm.com/computers?view=table", "doi": null, "title": "IBM Quantum Platform: Compute resources (table view)", "publisher": "IBM Quantum", "type": "vendor_technical" }, "measured_by": "IBM Quantum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Live value, rounded, no timestamp; snapshot read 27 Oct 2026. Roughly 6x Heron's 340K; the gain comes mostly from fast reset. IBM's Nighthawk r2 blog quotes throughput as over 100,000 circuits per second (a different measure, not CLOPS). Live value from IBM's calibration table, read on 27 Oct 2026; IBM updates it regularly.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000014/" }, { "id": "r-000015", "machine_id": "ibm-nighthawk-r2", "benchmark_id": "physical-qubit-count", "method_version": null, "comparison_group": "physical-qubit-count@gate-based", "value": 120, "value_text": "120 programmable qubits, 218 couplers, 458 physical quantum elements in total", "unit": "qubits", "uncertainty": null, "configuration": "ibm_phoenix. Square lattice, most qubits connected to four nearest neighbours; 218 tunable couplers plus 120 reset elements.", "measured_on": null, "published_on": { "date": "2026-08-31", "precision": "day" }, "source": { "url": "https://www.ibm.com/quantum/blog/nighthawk-r2", "doi": null, "title": "IBM Quantum Nighthawk r2 (IBM Quantum blog)", "publisher": "IBM Quantum (Holger Haas, David McKay, Robert Davis)", "type": "vendor_technical" }, "measured_by": "IBM Quantum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "IBM's platform table lists 'Physical qubits 458' for ibm_phoenix (and 332 for Heron), which counts couplers and reset elements, not computational qubits. Use 120 as the qubit count; do not compare 458 with other vendors' qubit counts.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000015/" }, { "id": "r-000016", "machine_id": "quantinuum-helios", "benchmark_id": "physical-qubit-count", "method_version": null, "comparison_group": "physical-qubit-count@gate-based", "value": 98, "value_text": "98-qubit trapped-ion quantum processor", "unit": "qubits", "uncertainty": null, "configuration": "Full Helios processor as described in the paper", "measured_on": null, "published_on": { "date": "2026-06-17", "precision": "day" }, "source": { "url": "https://www.nature.com/articles/s41586-026-10676-4", "doi": "10.1038/s41586-026-10676-4", "title": "A 98-qubit trapped-ion quantum computer with all-to-all connectivity", "publisher": "Nature (Springer Nature), vol. 655, pp. 81-86", "type": "peer_reviewed" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Qubit count is the hardware design figure stated by the builder; it does not say all 98 qubits are usable at the quoted fidelities in every program.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000016/" }, { "id": "r-000017", "machine_id": "quantinuum-helios", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Clifford two-qubit randomized benchmarking (2QRB), leakage accounted for; cycle benchmarking also used", "comparison_group": "two-qubit-gate-fidelity@mean", "value": 99.921, "value_text": "zone-averaged 2Q average infidelity 7.9(2) x 10^-4 (fidelity about 99.921%)", "unit": "%", "uncertainty": { "type": "other", "low": 99.919, "high": 99.923, "note": "converted from the stated error and its uncertainty" }, "configuration": "Average over 8 qubit pairs in the 8 operation zones", "measured_on": null, "published_on": { "date": "2026-06-17", "precision": "day" }, "source": { "url": "https://www.nature.com/articles/s41586-026-10676-4", "doi": "10.1038/s41586-026-10676-4", "title": "A 98-qubit trapped-ion quantum computer with all-to-all connectivity", "publisher": "Nature (Springer Nature), vol. 655, pp. 81-86", "type": "peer_reviewed" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Component-level figure averaged across the operation zones, not a full-system benchmark. Value is an infidelity as stated; the 99.921% fidelity is the arithmetic complement, not a separately quoted number in the fetched text. Converted from the stated average infidelity of 7.9(2)×10⁻⁴.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000017/" }, { "id": "r-000018", "machine_id": "quantinuum-helios", "benchmark_id": "logical-qubit-count", "method_version": "Two-level concatenated iceberg QEC code (I8 o I6, distance 4)", "comparison_group": "logical-qubit-count@error-corrected", "value": 48, "value_text": "up to 48 logical qubits (error-corrected)", "unit": "logical qubits", "uncertainty": null, "configuration": "Error correction with concatenated iceberg codes on the 98-qubit Helios processor", "measured_on": null, "published_on": { "date": "2026-02-25", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2602.22211", "doi": "10.48550/arXiv.2602.22211", "title": "Computing with many encoded logical qubits beyond break-even", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Preprint, not peer-reviewed. Logical qubits are code-specific (distance-4 iceberg concatenation); not comparable one-to-one with surface-code logical qubits. Physical and logical qubits are separate counts.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000018/" }, { "id": "r-000019", "machine_id": "quantinuum-helios", "benchmark_id": "logical-error-rate", "method_version": "QEC memory experiment, concatenated iceberg code I8 o I6 (distance 4)", "comparison_group": "logical-error-rate@concatenated-iceberg-d4-memory", "value": 4e-05, "value_text": "average QEC-cycle infidelity per logical qubit upper-bounded by 4 x 10^-5", "unit": "error per cycle", "uncertainty": { "type": "other", "low": 0, "high": 4e-05, "note": "Upper bound only: no logical errors recorded in 5,000 shots per basis" }, "configuration": "48 logical qubits, distance-4 concatenated iceberg code, memory experiment; compared against unencoded physical memory error of 1.1 x 10^-4", "measured_on": null, "published_on": { "date": "2026-02-25", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2602.22211", "doi": "10.48550/arXiv.2602.22211", "title": "Computing with many encoded logical qubits beyond break-even", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Preprint. An upper bound from zero observed failures, not a measured rate. Memory (idle) error only, not a logical gate error. Comparing across platforms needs care because codes and per-cycle definitions differ. Upper bound: no errors were seen in 5,000 shots per basis.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000019/" }, { "id": "r-000020", "machine_id": "quantinuum-h2", "benchmark_id": "quantum-volume", "method_version": "Quantum Volume (heavy output probability > 2/3 test)", "comparison_group": "quantum-volume@standard", "value": 33554432, "value_text": "33.5 million (2^25)", "unit": "quantum volume", "uncertainty": null, "configuration": "H2-2. 25-qubit QV circuits on H2-2 (raw data file n25_H2-2_raw_results.json in Quantinuum's public QV data repository)", "measured_on": null, "published_on": { "date": "2025-11-03", "precision": "day" }, "source": { "url": "https://docs.quantinuum.com/product-updates/posts/systems_11_2025/", "doi": null, "title": "Quantinuum Systems Releases November 3rd, 2025", "publisher": "Quantinuum", "type": "vendor_technical" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Company-reported. The fetched page states '33.5 million' rather than an exact figure; 2^25 = 33,554,432 is inferred from the n25 data file name. Heavy-output probability and confidence interval were not read. Measurement date not stated (trade press reported an announcement in Sept 2025, not used as a source).", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000020/" }, { "id": "r-000021", "machine_id": "quantinuum-h2", "benchmark_id": "quantum-volume", "method_version": "Quantum Volume (heavy output probability > 2/3 test)", "comparison_group": "quantum-volume@standard", "value": 8388608, "value_text": "2^23 = 8,388,608", "unit": "quantum volume", "uncertainty": null, "configuration": "H2-1. 23-qubit QV circuits (raw data file n23_H2-1_raw_results.json in Quantinuum's public QV data repository)", "measured_on": null, "published_on": { "date": "2025-05-12", "precision": "day" }, "source": { "url": "https://www.quantinuum.com/blog/quantum-volume-milestone", "doi": null, "title": "Quantinuum Dominates the Quantum Landscape: New World-Record in Quantum Volume", "publisher": "Quantinuum", "type": "vendor_technical" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "C", "limitations": "Company blog post. The post names 'System Model H2' only; the H2-1 unit comes from the data file name. Heavy-output probability and confidence interval not read. Superseded by the 2^25 result on H2-2. Quantinuum publishes the raw data for its quantum volume tests.", "ranking_eligible": false, "eligibility_note": "Grade C: announcement-level evidence is shown with a warning and never ranked.", "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000021/" }, { "id": "r-000022", "machine_id": "quantinuum-h2", "benchmark_id": "physical-qubit-count", "method_version": null, "comparison_group": "physical-qubit-count@gate-based", "value": 56, "value_text": "up to 56 qubits", "unit": "qubits", "uncertainty": null, "configuration": "56-qubit configuration. H2 trapped-ion processor as used for random circuit sampling experiments", "measured_on": null, "published_on": { "date": "2025", "precision": "year" }, "source": { "url": "https://arxiv.org/abs/2406.02501", "doi": "10.1103/PhysRevX.15.021052", "title": "The computational power of random quantum circuits in arbitrary geometries", "publisher": "Physical Review X 15, 021052 (2025) (arXiv:2406.02501)", "type": "peer_reviewed" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "The journal page itself could not be fetched (HTTP 403); value read from the arXiv abstract page, which lists the PRX journal reference.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000022/" }, { "id": "r-000023", "machine_id": "quantinuum-h2", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Two-qubit Clifford randomized benchmarking, Clifford error scaled to per native UZZ(pi/2) gate", "comparison_group": "two-qubit-gate-fidelity@mean", "value": 99.843, "value_text": "99.843(5)% two-qubit gate fidelity (per-gate error 1.57(5) x 10^-3)", "unit": "%", "uncertainty": { "type": "other", "low": 99.838, "high": 99.848, "note": "Stated as 99.843(5)%" }, "configuration": "56-qubit configuration. 2Q RB on the eight gate-zone qubits with standard Clifford randomization", "measured_on": null, "published_on": { "date": "2025", "precision": "year" }, "source": { "url": "https://arxiv.org/abs/2406.02501", "doi": "10.1103/PhysRevX.15.021052", "title": "The computational power of random quantum circuits in arbitrary geometries", "publisher": "Physical Review X 15, 021052 (2025) (arXiv:2406.02501)", "type": "peer_reviewed" }, "measured_by": "Quantinuum", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Component-level figure from the gate zones, not a full-system benchmark. Journal page returned 403; value read from arXiv abstract and HTML full text.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000023/" }, { "id": "r-000024", "machine_id": "quantinuum-h2", "benchmark_id": "logical-error-rate", "method_version": "Carbon [[12,2,4]] CSS code, logical Bell-state preparation with pre- and post-selection", "comparison_group": "logical-error-rate@carbon-code-bell-postselected", "value": 1e-05, "value_text": "0.001% logical error rate (vs 0.8% physical), about 800x lower", "unit": "error per circuit", "uncertainty": null, "configuration": "Two pairs of logical qubits (4 logical qubits) entangled with the [[12,2,4]] code; pre-selection kept 15,483 of 22,000 runs, post-selection kept 15,409", "measured_on": null, "published_on": { "date": "2024-04-02", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2404.02280", "doi": "10.48550/arXiv.2404.02280", "title": "Demonstration of logical qubits and repeated error correction with better-than-physical error rates", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Microsoft and Quantinuum (joint author team)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Relies on post-selection (error detection); about 30% of runs discarded. With pre-selection only, logical error rate was 0.17% (4.7x lower). Single-circuit error rate, not a per-cycle rate. Preprint; no journal reference listed on arXiv. This is an error per whole circuit with pre- and post-selection, not per correction cycle.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000024/" }, { "id": "r-000025", "machine_id": "ionq-tempo", "benchmark_id": "algorithmic-qubits", "method_version": "#AQ (IonQ algorithmic qubit benchmark suite)", "comparison_group": "algorithmic-qubits@ionq-aq", "value": 64, "value_text": "#AQ 64", "unit": "algorithmic qubits", "uncertainty": null, "configuration": "Achieved on an IonQ Tempo system", "measured_on": null, "published_on": { "date": "2025-09-25", "precision": "day" }, "source": { "url": "https://www.ionq.com/news/ionq-achieves-record-breaking-quantum-performance-milestone-of-aq-64", "doi": null, "title": "IonQ Achieves Record Breaking Quantum Performance Milestone of #AQ 64", "publisher": "IonQ", "type": "announcement" }, "measured_by": "IonQ", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "C", "limitations": "Company press release; no data, circuit details or fidelity threshold given on the page. #AQ is an IonQ-defined benchmark and is not directly comparable with Quantum Volume.", "ranking_eligible": false, "eligibility_note": "Grade C: announcement-level evidence is shown with a warning and never ranked.", "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000025/" }, { "id": "r-000026", "machine_id": "ionq-tempo", "benchmark_id": "physical-qubit-count", "method_version": null, "comparison_group": "physical-qubit-count@gate-based", "value": 100, "value_text": "100 qubits", "unit": "qubits", "uncertainty": null, "configuration": "Vendor specification", "measured_on": null, "published_on": null, "source": { "url": "https://www.ionq.com/quantum-systems/compare", "doi": null, "title": "IonQ quantum systems comparison", "publisher": "IonQ", "type": "vendor_technical" }, "measured_by": "IonQ", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Vendor marketing specification; page undated. Same table lists Tempo system availability as 'Late 2026'.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000026/" }, { "id": "r-000027", "machine_id": "ionq-tempo", "benchmark_id": "two-qubit-gate-fidelity", "method_version": null, "comparison_group": "two-qubit-gate-fidelity@statistic-not-stated", "value": 99.9, "value_text": "99.9%", "unit": "%", "uncertainty": null, "configuration": "Vendor specification table, row '2QG fidelity'", "measured_on": null, "published_on": null, "source": { "url": "https://www.ionq.com/quantum-systems/compare", "doi": null, "title": "IonQ quantum systems comparison", "publisher": "IonQ", "type": "vendor_technical" }, "measured_by": "IonQ", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "C", "limitations": "Vendor spec table with no method, statistic (median/average), date or uncertainty. Treat as headline claim only. No method or statistic is stated.", "ranking_eligible": false, "eligibility_note": "Grade C: announcement-level evidence is shown with a warning and never ranked.", "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000027/" }, { "id": "r-000028", "machine_id": "ionq-forte", "benchmark_id": "algorithmic-qubits", "method_version": "#AQ (IonQ algorithmic qubit benchmark suite)", "comparison_group": "algorithmic-qubits@ionq-aq", "value": 36, "value_text": "#AQ 36", "unit": "algorithmic qubits", "uncertainty": null, "configuration": "Vendor specification for the 36-qubit Forte", "measured_on": null, "published_on": null, "source": { "url": "https://www.ionq.com/quantum-systems/compare", "doi": null, "title": "IonQ quantum systems comparison", "publisher": "IonQ", "type": "vendor_technical" }, "measured_by": "IonQ", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "C", "limitations": "Vendor page, undated, no data or method details. #AQ is IonQ-defined. Specification-page figure with no method or date.", "ranking_eligible": false, "eligibility_note": "Grade C: announcement-level evidence is shown with a warning and never ranked.", "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000028/" }, { "id": "r-000029", "machine_id": "ionq-forte", "benchmark_id": "algorithmic-qubits", "method_version": "#AQ (algorithmic qubit benchmark suite)", "comparison_group": "algorithmic-qubits@ionq-aq", "value": 29, "value_text": "#AQ 29", "unit": "algorithmic qubits", "uncertainty": null, "configuration": "IonQ Forte configured as a single-chain 30-qubit trapped-ion computer", "measured_on": null, "published_on": { "date": "2024-11-07", "precision": "day" }, "source": { "url": "https://quantum-journal.org/papers/q-2024-11-07-1516/", "doi": "10.22331/q-2024-11-07-1516", "title": "Benchmarking a trapped-ion quantum computer with 30 qubits", "publisher": "Quantum 8, 1516 (2024)", "type": "peer_reviewed" }, "measured_by": "IonQ", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Earlier 30-qubit configuration of Forte; superseded by the vendor-stated #AQ 36. Authors are IonQ staff.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000029/" }, { "id": "r-000030", "machine_id": "ionq-forte", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Direct randomized benchmarking (DRB) on all 435 qubit pairs", "comparison_group": "two-qubit-gate-fidelity@median", "value": 99.536, "value_text": "median two-qubit error rate 46.4 x 10^-4 across 435 pairs (fidelity about 99.54%)", "unit": "%", "uncertainty": null, "configuration": "DRB across all 30 choose 2 = 435 gate pairs, each pair measured at least 4 times over about six months", "measured_on": null, "published_on": { "date": "2024-11-07", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2308.05071", "doi": "10.22331/q-2024-11-07-1516", "title": "Benchmarking a trapped-ion quantum computer with 30 qubits", "publisher": "Quantum 8, 1516 (2024) (arXiv:2308.05071)", "type": "peer_reviewed" }, "measured_by": "IonQ", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Value is a median error rate as stated; 99.54% fidelity is a conversion, not quoted. Earlier 30-qubit configuration; current vendor spec says 99.6%. Measurement window (about six months) has no calendar dates. Converted from the stated median error of 46.4×10⁻⁴.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000030/" }, { "id": "r-000031", "machine_id": "ionq-oxford-ionics-prototype", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Subspace Leakage Error Randomized Benchmarking (SLERB), 'smooth gate' without ground-state cooling", "comparison_group": "two-qubit-gate-fidelity@single-pair", "value": 99.9916, "value_text": "estimated two-qubit gate error 8.4(7) x 10^-5 (fidelity >99.99%)", "unit": "%", "uncertainty": { "type": "other", "low": 99.9909, "high": 99.9923, "note": "converted from the stated error and its uncertainty" }, "configuration": "Two-qubit test setup, sequences up to N=200 Cliffords; SPAM assumed negligible (no y-offset in fit); above the Doppler limit", "measured_on": null, "published_on": { "date": "2025-10-20", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2510.17286", "doi": "10.48550/arXiv.2510.17286", "title": "Trapped-ion two-qubit gates with >99.99% fidelity without ground-state cooling", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Oxford Ionics (an IonQ company)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Two-qubit lab prototype, not a product system. Preprint. Described as an 'estimated' error; SPAM assumed negligible. Not comparable to system-wide averages on full machines. Converted from the stated error of 8.4(7)×10⁻⁵.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000031/" }, { "id": "r-000032", "machine_id": "quera-harvard-mit-array", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Global randomized benchmarking (random global single-qubit rotations between CZ gates); cross-checked by repeated-CZ Bell-state method", "comparison_group": "two-qubit-gate-fidelity@mean", "value": 99.54, "value_text": "FCZ = 99.54(2)% (time-optimal CZ, 20 atoms / 10 gate sites in parallel)", "unit": "%", "uncertainty": { "type": "other", "low": 99.52, "high": 99.56, "note": "Stated as 99.54(2)%; paper says error bars are 68% confidence intervals" }, "configuration": "CZ gates on 10 pairs (20 atoms) in parallel, up to 20 CZ gates with random global single-qubit rotations. Same gate on 60 atoms in parallel gave 99.48(2)%; smooth-amplitude gate 99.55(3)%; Bell-state method 99.52(4)%.", "measured_on": null, "published_on": { "date": "2023-10-11", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2304.05420", "doi": "10.1038/s41586-023-06481-y", "title": "High-fidelity parallel entangling gates on a neutral atom quantum computer", "publisher": "Nature 622, 268-272 (2023); preprint arXiv:2304.05420", "type": "peer_reviewed" }, "measured_by": "Harvard University, MIT and QuEra Computing (Lukin group collaboration)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Research system, not a commercial QuEra product. Benchmarking uses only global rotations, which the authors note makes it less sensitive to some error types. Gate fidelity, not a whole-system benchmark.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000032/" }, { "id": "r-000033", "machine_id": "quera-harvard-mit-array", "benchmark_id": "logical-qubit-count", "method_version": null, "comparison_group": "logical-qubit-count@error-detecting", "value": 48, "value_text": "up to 48 logical qubits in [[8,3,2]] code blocks (128 physical qubits for this circuit)", "unit": "logical qubits", "uncertainty": null, "configuration": "3D [[8,3,2]] error-detecting code (distance 2), 16 blocks of 3 logical qubits; IQP sampling circuit with 228 logical two-qubit gates and 48 logical CCZ gates; results with error detection (post-selection). XEB of about 0.1 reported for the 48-logical-qubit circuit.", "measured_on": null, "published_on": { "date": "2023-12-06", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2312.03982", "doi": "10.1038/s41586-023-06927-3", "title": "Logical quantum processor based on reconfigurable atom arrays", "publisher": "Nature 626, 58-65 (2024); preprint arXiv:2312.03982", "type": "peer_reviewed" }, "measured_by": "Harvard University, MIT, QuEra Computing and NIST/University of Maryland (Lukin group collaboration)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "The [[8,3,2]] code is distance 2: it detects errors but cannot correct them, and results rely on post-selection. No per-logical-qubit error rate is given for the 48-qubit circuit; performance is reported as an XEB score (about 0.1). Research system, not a commercial product.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000033/" }, { "id": "r-000035", "machine_id": "atom-computing-ytterbium-array", "benchmark_id": "logical-qubit-count", "method_version": null, "comparison_group": "logical-qubit-count@error-detecting", "value": 24, "value_text": "24 logical qubits in a cat state, encoded into 48 atoms with the [[4,2,2]] code", "unit": "logical qubits", "uncertainty": null, "configuration": "256-qubit processor, with Microsoft. [[4,2,2]] error-detecting code (distance 2), 12 code blocks; encoded circuit used 59 atoms and 101 physical CZ gates; error detection plus correction of lost atoms (on average 1.8). Total X+Z error pX + pZ = 26.6% encoded vs 42.0% unencoded baseline. Same paper: Bernstein-Vazirani on up to 28 logical qubits in 112 atoms ([[4,1,2]] code).", "measured_on": null, "published_on": { "date": "2024-11-18", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2411.11822", "doi": "10.48550/arXiv.2411.11822", "title": "Fault-tolerant quantum computation with a neutral atom processor", "publisher": "arXiv (preprint, v3 9 Jun 2025)", "type": "preprint" }, "measured_by": "Microsoft Quantum and Atom Computing", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Distance-2 code: detects errors (discarding those shots) and corrects only atom loss. The encoded cat state still has a combined X+Z error of 26.6%, so these are not low-error logical qubits. Not peer-reviewed (no journal reference on arXiv).", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000035/" }, { "id": "r-000036", "machine_id": "atom-computing-ytterbium-array", "benchmark_id": "physical-qubit-count", "method_version": null, "comparison_group": "physical-qubit-count@gate-based", "value": 256, "value_text": "quantum processor with 256 qubits", "unit": "qubits", "uncertainty": null, "configuration": "256-qubit processor, with Microsoft. Processor used for the logical-qubit experiments in the same paper; layouts of 250 and 256 atoms are shown for code-block preparation.", "measured_on": null, "published_on": { "date": "2024-11-18", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2411.11822", "doi": "10.48550/arXiv.2411.11822", "title": "Fault-tolerant quantum computation with a neutral atom processor", "publisher": "arXiv (preprint, v3 9 Jun 2025)", "type": "preprint" }, "measured_by": "Microsoft Quantum and Atom Computing", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Qubit count is a capacity figure, not a quality measure. Not peer-reviewed.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000036/" }, { "id": "r-000037", "machine_id": "atom-computing-ytterbium-array", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Two-qubit Clifford randomized benchmarking (plus symmetric subspace RB), depolarizing-model extraction of CZ fidelity", "comparison_group": "two-qubit-gate-fidelity@single-pair", "value": 99.4, "value_text": "CZ fidelity 99.40(3)% without post-selection; 99.72(3)% with post-selection", "unit": "%", "uncertainty": { "type": "other", "low": 99.37, "high": 99.43, "note": "Stated as 99.40(3)% (without post-selection)" }, "configuration": "Circuits with more than 200 CZ gates applied to one or two pairs of atoms. Single-qubit Clifford RB fidelity 99.963(2)%.", "measured_on": null, "published_on": { "date": "2025-05-21", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2411.11708", "doi": "10.1103/PRXQuantum.6.020334", "title": "High-fidelity universal gates in the 171Yb ground state nuclear spin qubit", "publisher": "PRX Quantum 6, 020334 (2025); preprint arXiv:2411.11708", "type": "peer_reviewed" }, "measured_by": "Atom Computing", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Measured on one or two atom pairs, not across a full array. The higher 99.72% figure requires post-selection (discarding atom-loss events); the 99.40% figure is without post-selection.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000037/" }, { "id": "r-000038", "machine_id": "infleqtion-cesium-array", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Repeated-CZ benchmarking of two-atom survival and cycle polarization, combined with a model-based leakage estimate", "comparison_group": "two-qubit-gate-fidelity@statistic-not-stated", "value": 99.35, "value_text": "CZ fidelity 99.35(4)%; 99.73(3)% when excluding atom-loss events", "unit": "%", "uncertainty": { "type": "other", "low": 99.31, "high": 99.39, "note": "Stated as 99.35(4)%" }, "configuration": "24-qubit computational array. Nearest-neighbour CZ gates via individual optical addressing (no atom shuttling). Local single-qubit RZ fidelity 99.902(8)%.", "measured_on": null, "published_on": { "date": "2025-08-25", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2408.08288", "doi": "10.1103/66s8-jj18", "title": "A universal neutral-atom quantum computer with individual optical addressing and non-destructive readout", "publisher": "PRX Quantum 6, 030334 (2025); preprint arXiv:2408.08288", "type": "peer_reviewed" }, "measured_by": "Infleqtion (with University of Wisconsin-Madison co-author)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Fidelity partly relies on a model-based leakage estimate rather than full randomized benchmarking. The 99.73% figure excludes atom-loss events. The paper does not name the system as a commercial product.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000038/" }, { "id": "r-000039", "machine_id": "infleqtion-cesium-array", "benchmark_id": "logical-qubit-count", "method_version": null, "comparison_group": "logical-qubit-count@error-detecting", "value": 12, "value_text": "12 logical qubits in a constant-depth logical CX ladder, [[4,2,2]] code", "unit": "logical qubits", "uncertainty": null, "configuration": "Sqale QPU, 114 qubits. [[4,2,2]] error-detecting code (distance 2); constant-depth CX ladder on 12 logical qubits; encoded error rate more than 2x lower than unencoded (8-logical-qubit version nearly 4x lower). Post-processing survival rate 0.009 encoded vs 0.768 unencoded.", "measured_on": null, "published_on": { "date": "2025-09-16", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2509.13247", "doi": "10.48550/arXiv.2509.13247", "title": "Demonstration of a Logical Architecture Uniting Motion and In-Place Entanglement", "publisher": "arXiv (preprint, v2 2 Apr 2026)", "type": "preprint" }, "measured_by": "Infleqtion (with University of Chicago and University of Wisconsin-Madison co-authors)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Distance-2 error detection with heavy post-selection: only 0.9% of encoded 12-logical-qubit shots survived post-processing. No absolute logical error rate per qubit is given in the abstract, only a relative improvement. Not peer-reviewed.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000039/" }, { "id": "r-000040", "machine_id": "rigetti-ankaa-3", "benchmark_id": "two-qubit-gate-fidelity", "method_version": null, "comparison_group": "two-qubit-gate-fidelity@median", "value": 99.5, "value_text": "99.5% median two-qubit gate fidelity", "unit": "%", "uncertainty": null, "configuration": "Median across the device; gate type and measurement method not stated on the fetched page.", "measured_on": null, "published_on": { "date": "2024-12-23", "precision": "day" }, "source": { "url": "https://www.rigetti.com/news/rigetti-computing-launches-84-qubit-ankaa-3-system-achieves-99-5-median-two-qubit-gate-fidelity-milestone", "doi": null, "title": "Rigetti Computing launches 84-qubit Ankaa-3 system; Achieves 99.5% median two-qubit gate fidelity milestone", "publisher": "Rigetti Computing", "type": "announcement" }, "measured_by": "Rigetti Computing", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "C", "limitations": "Company announcement with no method or data. Search snippets of the full press release (not fetched, 403) indicate the 99.5% applies to fSim gates while the standard iSWAP gate median was 99.0%; treat 99.5% as the best gate type, not the general-purpose gate.", "ranking_eligible": false, "eligibility_note": "Grade C: announcement-level evidence is shown with a warning and never ranked.", "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000040/" }, { "id": "r-000041", "machine_id": "rigetti-cepheus-1-108q", "benchmark_id": "two-qubit-gate-fidelity", "method_version": null, "comparison_group": "two-qubit-gate-fidelity@median", "value": 98.82, "value_text": "Two-qubit gates, CZ (median fidelity, per op.): 98.82%", "unit": "%", "uncertainty": null, "configuration": "Median CZ fidelity per operation across the device, from Rigetti's live 'Rigetti Right Now' panel (deployed 04.07.26; 107 qubits; median single-qubit fidelity 99.85%; median T1 29 and T2 11, units shown as microseconds in header).", "measured_on": null, "published_on": null, "source": { "url": "https://www.rigetti.com/", "doi": null, "title": "Rigetti Computing home page, 'Rigetti Right Now' specification panel", "publisher": "Rigetti Computing", "type": "vendor_technical" }, "measured_by": "Rigetti Computing", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Live figure with no measurement date or method; page read on 27 Oct 2026 and may change. Lower than the 99.1% median Rigetti cited at general availability (April 2026), which was not read on a Rigetti page.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000041/" }, { "id": "r-000042", "machine_id": "iqm-garnet", "benchmark_id": "quantum-volume", "method_version": "Quantum volume (IBM protocol): heavy-output probability > 2/3 within two standard deviations", "comparison_group": "quantum-volume@standard", "value": 32, "value_text": "QV = 2^5 = 32", "unit": "quantum volume", "uncertainty": null, "configuration": "Square circuits of random SU(4) layers; passed at width/depth 5.", "measured_on": null, "published_on": { "date": "2024-08-22", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2408.12433", "doi": "10.48550/arXiv.2408.12433", "title": "Technology and Performance Benchmarks of IQM's 20-Qubit Quantum Computer", "publisher": "arXiv (preprint; also issued as IQM white paper)", "type": "preprint" }, "measured_by": "IQM Quantum Computers", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Company-run, not peer-reviewed. No measurement date given.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000042/" }, { "id": "r-000043", "machine_id": "iqm-garnet", "benchmark_id": "clops", "method_version": "Original (virtual) CLOPS: M = 100 templates, K = 10 parameter updates, S = 100 shots, D = log2 QV layers", "comparison_group": "clops@original-virtual", "value": 2600, "value_text": "CLOPS of 2600", "unit": "layers per second", "uncertainty": null, "configuration": "QV-style circuits with D = log2(QV) = 5 layers.", "measured_on": null, "published_on": { "date": "2024-08-22", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2408.12433", "doi": "10.48550/arXiv.2408.12433", "title": "Technology and Performance Benchmarks of IQM's 20-Qubit Quantum Computer", "publisher": "arXiv (preprint; also issued as IQM white paper)", "type": "preprint" }, "measured_by": "IQM Quantum Computers", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "The paper calls this the 'virtual CLOPS' (original definition), not the newer hardware-aware CLOPS_h, so it is not directly comparable with IBM's current CLOPS figures.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000043/" }, { "id": "r-000044", "machine_id": "iqm-garnet", "benchmark_id": "two-qubit-gate-fidelity", "method_version": "Interleaved randomized benchmarking of CZ gates, run in parallel in distance-two groups", "comparison_group": "two-qubit-gate-fidelity@median", "value": 99.5, "value_text": "median 2-qubit gate fidelity of 99.5% (median CZ error 5 x 10^-3)", "unit": "%", "uncertainty": null, "configuration": "Median across qubit pairs; gates characterized in parallel with pairs separated by two idling couplers and one idling qubit. Benchmarking in distance-one groups lowers mean CZ fidelity by 0.6 (as stated). Median single-qubit gate error 9 x 10^-4.", "measured_on": null, "published_on": { "date": "2024-08-22", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2408.12433", "doi": "10.48550/arXiv.2408.12433", "title": "Technology and Performance Benchmarks of IQM's 20-Qubit Quantum Computer", "publisher": "arXiv (preprint; also issued as IQM white paper)", "type": "preprint" }, "measured_by": "IQM Quantum Computers", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Measured with spatially separated pairs to limit crosstalk; fully parallel (distance-one) operation gives lower fidelity. Company-run, not peer-reviewed.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000044/" }, { "id": "r-000045", "machine_id": "iqm-garnet", "benchmark_id": "q-score", "method_version": "Atos Q-score (Max-Cut on Erdos-Renyi graphs, p = 0.5), depth p = 1 QAOA, pass threshold beta(n) >= 0.2, with one virtual node", "comparison_group": "q-score@gate-based-qaoa-p1", "value": 15, "value_text": "Q-score of 15", "unit": "Q-score", "uncertainty": null, "configuration": "Depth-1 QAOA; the virtual-node technique solves a size-n problem on n - 1 qubits and raises the score by 1.", "measured_on": null, "published_on": { "date": "2024-08-22", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2408.12433", "doi": "10.48550/arXiv.2408.12433", "title": "Technology and Performance Benchmarks of IQM's 20-Qubit Quantum Computer", "publisher": "arXiv (preprint; also issued as IQM white paper)", "type": "preprint" }, "measured_by": "IQM Quantum Computers", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Uses the virtual-node trick (score of 15 on 14 qubits), so not directly comparable with Q-scores obtained without it. Company-run, not peer-reviewed.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000045/" }, { "id": "r-000046", "machine_id": "iqm-emerald", "benchmark_id": "quantum-volume", "method_version": null, "comparison_group": "quantum-volume@standard", "value": 64, "value_text": "Quantum volume of 64", "unit": "quantum volume", "uncertainty": null, "configuration": "Not stated.", "measured_on": null, "published_on": { "date": "2025-08-05", "precision": "day" }, "source": { "url": "https://iqm.tech/blog/new-on-iqm-resonance-iqm-emerald-almost-3x-qubits-for-the-same-reliable-quality/", "doi": null, "title": "New on IQM Resonance: IQM Emerald - almost 3x qubits for the same reliable quality", "publisher": "IQM Quantum Computers", "type": "announcement" }, "measured_by": "IQM Quantum Computers", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "C", "limitations": "Company blog post with no method, data or measurement date.", "ranking_eligible": false, "eligibility_note": "Grade C: announcement-level evidence is shown with a warning and never ranked.", "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000046/" }, { "id": "r-000047", "machine_id": "iqm-emerald", "benchmark_id": "two-qubit-gate-fidelity", "method_version": null, "comparison_group": "two-qubit-gate-fidelity@median", "value": 99.5, "value_text": "median two-qubit gate fidelity 99.5%; maximum up to 99.85%", "unit": "%", "uncertainty": null, "configuration": "Median across qubit pairs (CZ gate); method not stated.", "measured_on": null, "published_on": { "date": "2025-08-05", "precision": "day" }, "source": { "url": "https://iqm.tech/blog/new-on-iqm-resonance-iqm-emerald-almost-3x-qubits-for-the-same-reliable-quality/", "doi": null, "title": "New on IQM Resonance: IQM Emerald - almost 3x qubits for the same reliable quality", "publisher": "IQM Quantum Computers", "type": "announcement" }, "measured_by": "IQM Quantum Computers", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "C", "limitations": "Company blog post with no method or measurement date; launch-time figure that may have changed.", "ranking_eligible": false, "eligibility_note": "Grade C: announcement-level evidence is shown with a warning and never ranked.", "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000047/" }, { "id": "r-000048", "machine_id": "quantinuum-helios", "benchmark_id": "quops", "method_version": "QUOPS v1 (arXiv:2609.12146v1), physical-qubit architecture", "comparison_group": "quops@v1-physical", "value": 1504, "value_text": "Q = 1504 at 303 QUOPS/s (width 16). With leakage postselection: 1504 at 344 QUOPS/s; 1824 at 247 QUOPS/s.", "unit": "QUOPS score", "uncertainty": null, "configuration": "Helios-1. Physical qubits, width w=16, passing shape (16,1504); 20 QUOPS circuits x 200 shots (4000 shots); timed run in reserved window, runtime 284 min (SI Table VI). Rate 303 QUOPS/s.", "measured_on": { "date": "2026-06-02", "precision": "day" }, "published_on": { "date": "2026-09-10", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2609.12146", "doi": "10.48550/arXiv.2609.12146", "title": "Benchmarking the computational power of quantum computers", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Quantinuum (co-authors of the Sandia-led paper)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "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.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000048/" }, { "id": "r-000049", "machine_id": "quantinuum-helios", "benchmark_id": "quops", "method_version": "QUOPS v1 (arXiv:2609.12146v1), logical-qubit architecture ([[7,1,3]] code)", "comparison_group": "quops@v1-logical", "value": 40, "value_text": "Q = 40 at 4.9 QUOPS/s (width 4, [[7,1,3]] code)", "unit": "QUOPS score", "uncertainty": null, "configuration": "Helios-1. Logical qubits: [[7,1,3]] (Steane) colour-code encoding, width w=4 logical qubits, size 40; rate 4.9 QUOPS/s (main-text Table I; SI Sec. VII C, Table V).", "measured_on": null, "published_on": { "date": "2026-09-10", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2609.12146", "doi": "10.48550/arXiv.2609.12146", "title": "Benchmarking the computational power of quantum computers", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Quantinuum (co-authors of the Sandia-led paper)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "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. Small logical demonstration (4 logical qubits); not comparable to physical-qubit scores without context.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000049/" }, { "id": "r-000050", "machine_id": "quantinuum-h2", "benchmark_id": "quops", "method_version": "QUOPS v1 (arXiv:2609.12146v1), physical-qubit architecture", "comparison_group": "quops@v1-physical", "value": 1320, "value_text": "Q = 1320 at 353 QUOPS/s (width 12). With leakage postselection: 1320 at 360 QUOPS/s; 1392 at 231 QUOPS/s.", "unit": "QUOPS score", "uncertainty": null, "configuration": "H2-1. Physical qubits, width w=12, passing shape (12,1320); 10 QUOPS circuits x 200 shots (2000 shots); timed run in reserved window, runtime 99 min 23 s (SI Table IV). Rate 353 QUOPS/s.", "measured_on": { "date": "2026-08-29", "precision": "day" }, "published_on": { "date": "2026-09-10", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2609.12146", "doi": "10.48550/arXiv.2609.12146", "title": "Benchmarking the computational power of quantum computers", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Quantinuum (co-authors of the Sandia-led paper)", "provenance": { "company_reported": true, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "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.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000050/" }, { "id": "r-000051", "machine_id": "google-willow", "benchmark_id": "quops", "method_version": "QUOPS v1 (arXiv:2609.12146v1), physical-qubit architecture", "comparison_group": "quops@v1-physical", "value": 216, "value_text": "Q = 216 at 2.0x10^7 QUOPS/s (width 6)", "unit": "QUOPS score", "uncertainty": null, "configuration": "Physical qubits, width w=6, shape (6,216); capability-region run: 100 QUOPS circuits per shape, 1000 shots each; rate run: 25 QUOPS circuits at (6,216) with 500k shots each. Compiled with cirq (CNOT->CZ, PhasedXZ). Rate 2.0x10^7 QUOPS/s.", "measured_on": null, "published_on": { "date": "2026-09-10", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2609.12146", "doi": "10.48550/arXiv.2609.12146", "title": "Benchmarking the computational power of quantum computers", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Ryan Byrne and Eliott Rosenberg (thanked in acknowledgments for running the Willow experiments; affiliation not stated in the paper)", "provenance": { "company_reported": false, "independent_measurement": false, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "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.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000051/" }, { "id": "r-000052", "machine_id": "ibm-heron-r3", "benchmark_id": "quops", "method_version": "QUOPS v1 (arXiv:2609.12146v1), physical-qubit architecture", "comparison_group": "quops@v1-physical", "value": 204, "value_text": "Q = 204 at 3.1x10^5 QUOPS/s (width 6)", "unit": "QUOPS score", "uncertainty": null, "configuration": "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.", "measured_on": null, "published_on": { "date": "2026-09-10", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2609.12146", "doi": "10.48550/arXiv.2609.12146", "title": "Benchmarking the computational power of quantum computers", "publisher": "arXiv", "type": "preprint" }, "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", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "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.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000052/" }, { "id": "r-000053", "machine_id": "quantinuum-h1", "benchmark_id": "quantum-volume", "method_version": "Quantum volume protocol (heavy-output probability > 2/3, z-confidence > 0.99), independent user-side implementation", "comparison_group": "quantum-volume@standard", "value": 512, "value_text": "QV >= 512 (lower bound; n = 9 passed, larger not completed)", "unit": "quantum volume", "uncertainty": null, "configuration": "H1-2. Black-box submission of OpenQASM via Quantinuum (Honeywell) API with default compiler optimisation; n=5 at 100 shots/circuit, n=6,8,9 at 20 shots/circuit. Access: Oak Ridge National Laboratory OLCF program.", "measured_on": null, "published_on": { "date": "2022", "precision": "year" }, "source": { "url": "https://arxiv.org/abs/2203.03816", "doi": "10.1109/TQE.2022.3184764", "title": "Quantum Volume in Practice: What Users Can Expect from NISQ Devices", "publisher": "IEEE Transactions on Quantum Engineering (vol. 3, 2022, art. 3102119); preprint arXiv:2203.03816", "type": "peer_reviewed" }, "measured_by": "Los Alamos National Laboratory", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "A", "limitations": "Measured as an external cloud user without white-box access; authors note QV results depend strongly on compilation effort and drift over time. Devices from 2022 are largely retired. Exact run dates not given.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000053/" }, { "id": "r-000054", "machine_id": "ionq-harmony", "benchmark_id": "quantum-volume", "method_version": "Quantum volume protocol (heavy-output probability > 2/3, z-confidence > 0.99), independent user-side implementation", "comparison_group": "quantum-volume@standard", "value": 8, "value_text": null, "unit": "quantum volume", "uncertainty": null, "configuration": "Qiskit-transpiled QV circuits submitted via Amazon Braket; passed n=3 after ~380 circuits, failed n=4 after 500 circuits. Access: Amazon Braket.", "measured_on": null, "published_on": { "date": "2022", "precision": "year" }, "source": { "url": "https://arxiv.org/abs/2203.03816", "doi": "10.1109/TQE.2022.3184764", "title": "Quantum Volume in Practice: What Users Can Expect from NISQ Devices", "publisher": "IEEE Transactions on Quantum Engineering (vol. 3, 2022, art. 3102119); preprint arXiv:2203.03816", "type": "peer_reviewed" }, "measured_by": "Los Alamos National Laboratory", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "A", "limitations": "Measured as an external cloud user without white-box access; authors note QV results depend strongly on compilation effort and drift over time. Devices from 2022 are largely retired. Exact run dates not given.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000054/" }, { "id": "r-000055", "machine_id": "rigetti-aspen-m-1", "benchmark_id": "quantum-volume", "method_version": "Quantum volume protocol (heavy-output probability > 2/3, z-confidence > 0.99), independent user-side implementation", "comparison_group": "quantum-volume@standard", "value": 8, "value_text": "QV 8 on 1 of 26 tested 3-qubit subsets (black-box QV < 2)", "unit": "quantum volume", "uncertainty": null, "configuration": "Qubit-subset enumeration on Rigetti QCS (quilc compilation); at n=3, 26 of 184 connected subsets tested, one passed at 99% z-confidence. Black-box run via Amazon Braket failed n=2. Access: Rigetti Quantum Cloud Services (QCS) and Amazon Braket.", "measured_on": null, "published_on": { "date": "2022", "precision": "year" }, "source": { "url": "https://arxiv.org/abs/2203.03816", "doi": "10.1109/TQE.2022.3184764", "title": "Quantum Volume in Practice: What Users Can Expect from NISQ Devices", "publisher": "IEEE Transactions on Quantum Engineering (vol. 3, 2022, art. 3102119); preprint arXiv:2203.03816", "type": "peer_reviewed" }, "measured_by": "Los Alamos National Laboratory", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": true, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "A", "limitations": "Measured as an external cloud user without white-box access; authors note QV results depend strongly on compilation effort and drift over time. Devices from 2022 are largely retired. Exact run dates not given.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000055/" }, { "id": "r-000056", "machine_id": "d-wave-advantage", "benchmark_id": "q-score", "method_version": "Atos Q-score (Max-Cut), adapted to quantum annealing; D-Wave default methods, 60 s time limit", "comparison_group": "q-score@annealer", "value": 140, "value_text": null, "unit": "Q-score", "uncertainty": null, "configuration": "Max-Cut QUBOs on Erdos-Renyi graphs of increasing size N, solved with D-Wave's standard out-of-the-box methods, 60 s time limit.", "measured_on": null, "published_on": { "date": "2022-08-16", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2208.07633", "doi": "10.1109/QSW55613.2022.00017", "title": "Evaluating the Q-score of Quantum Annealers", "publisher": "arXiv preprint; published in 2022 IEEE International Conference on Quantum Software (QSW) proceedings", "type": "preprint" }, "measured_by": "TNO (Netherlands Organisation for Applied Scientific Research)", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Q-score was defined by Atos for gate-based QAOA; here applied to an annealer using D-Wave's out-of-the-box methods with a 60 s time limit, so not directly comparable with gate-based Q-scores. Conference paper (peer-reviewed proceedings, not a journal). Run dates not given.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000056/" }, { "id": "r-000057", "machine_id": "d-wave-2000q", "benchmark_id": "q-score", "method_version": "Atos Q-score (Max-Cut), adapted to quantum annealing; D-Wave default methods, 60 s time limit", "comparison_group": "q-score@annealer", "value": 70, "value_text": null, "unit": "Q-score", "uncertainty": null, "configuration": "Max-Cut QUBOs on Erdos-Renyi graphs of increasing size N, solved with D-Wave's standard out-of-the-box methods, 60 s time limit.", "measured_on": null, "published_on": { "date": "2022-08-16", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2208.07633", "doi": "10.1109/QSW55613.2022.00017", "title": "Evaluating the Q-score of Quantum Annealers", "publisher": "arXiv preprint; published in 2022 IEEE International Conference on Quantum Software (QSW) proceedings", "type": "preprint" }, "measured_by": "TNO (Netherlands Organisation for Applied Scientific Research)", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Q-score was defined by Atos for gate-based QAOA; here applied to an annealer using D-Wave's out-of-the-box methods with a 60 s time limit, so not directly comparable with gate-based Q-scores. Conference paper (peer-reviewed proceedings, not a journal). Run dates not given.", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000057/" }, { "id": "r-000058", "machine_id": "ibm-heron-r3", "benchmark_id": "clops", "method_version": "Metriq-gym CLOPS = L*M*S / T_tot, twirled mode (not IBM's CLOPS_h definition)", "comparison_group": "clops@metriq-gym", "value": 360573, "value_text": null, "unit": "layers per second", "uncertainty": null, "configuration": "ibm_boston. N=100 qubits, L=100 layers, M=1000 circuits, S=100 shots, two-qubit gate cz; T_tot from start of first circuit to end of last (intended to exclude queueing).", "measured_on": null, "published_on": { "date": "2026-03-09", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2603.08680", "doi": null, "title": "Metriq: A Collaborative Platform for Benchmarking Quantum Computers", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Unitary Foundation (Metriq project)", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Preprint, not peer reviewed. Benchmark implementations are Metriq's own (metriq-gym) and may differ from vendor definitions. CLOPS definition and parameters differ from IBM's published CLOPS_h, so values are not directly comparable to IBM's figures. No per-device run date (overall data window Mar 2025 to Mar 2026).", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000058/" }, { "id": "r-000059", "machine_id": "ibm-heron-r3", "benchmark_id": "layer-fidelity", "method_version": "EPLG (error per layered gate, derived from layer fidelity) at chain length 100, metriq-gym implementation; closest ID is layer-fidelity", "comparison_group": "layer-fidelity@eplg-100-metriq", "value": 0.00308, "value_text": "EPLG-100 = 0.00308 (x10^-3 in source)", "unit": "error per layered gate", "uncertainty": null, "configuration": "ibm_boston. Layer-fidelity/EPLG on qubit chains of length 10, 20, 50, 100 where supported; value shown is chain length 100. Data captured November 2025 to February 2026.", "measured_on": null, "published_on": { "date": "2026-03-09", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2603.08680", "doi": null, "title": "Metriq: A Collaborative Platform for Benchmarking Quantum Computers", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Unitary Foundation (Metriq project)", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Preprint, not peer reviewed. Benchmark implementations are Metriq's own (metriq-gym) and may differ from vendor definitions. EPLG is an error rate (lower is better), not a layer fidelity; chain selection method not read in full. Measurement window only (Nov 2025 to Feb 2026).", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000059/" }, { "id": "r-000060", "machine_id": "quantinuum-h2", "benchmark_id": "layer-fidelity", "method_version": "EPLG (error per layered gate, derived from layer fidelity) at chain length 50, metriq-gym implementation; closest ID is layer-fidelity", "comparison_group": "layer-fidelity@eplg-50-metriq", "value": 0.00173, "value_text": "EPLG-50 = 0.00173 (x10^-3 in source)", "unit": "error per layered gate", "uncertainty": null, "configuration": "H2-2. Layer-fidelity/EPLG on qubit chains of length 10, 20, 50, 100 where supported; value shown is chain length 50. Data captured November 2025 to February 2026.", "measured_on": null, "published_on": { "date": "2026-03-09", "precision": "day" }, "source": { "url": "https://arxiv.org/abs/2603.08680", "doi": null, "title": "Metriq: A Collaborative Platform for Benchmarking Quantum Computers", "publisher": "arXiv", "type": "preprint" }, "measured_by": "Unitary Foundation (Metriq project)", "provenance": { "company_reported": false, "independent_measurement": true, "peer_reviewed": false, "independently_reproduced": false, "community_submitted": false, "audited": false }, "grade": "B", "limitations": "Preprint, not peer reviewed. Benchmark implementations are Metriq's own (metriq-gym) and may differ from vendor definitions. EPLG is an error rate (lower is better), not a layer fidelity; chain selection method not read in full. Measurement window only (Nov 2025 to Feb 2026).", "ranking_eligible": true, "eligibility_note": null, "status": "active", "superseded_by": null, "corrects": null, "history": [ { "date": "2026-10-27", "change": "added", "note": null } ], "added_on": "2026-10-27", "updated_on": "2026-10-27", "page_url": "https://qubitbenchmark.com/results/r-000060/" } ]