The result
The Quantum Optimization Benchmarking Library, published in Nature Computational Science in June 2026, defines ten classes of optimization problems and releases the instances, data and code openly [1][2].
What it establishes
Optimization is where commercial expectations of quantum computing are highest and where claims have been hardest to verify. Results have often been reported on problems chosen by the authors, against classical baselines of uneven strength. An independent 2025 study found that a hybrid annealing solver did well on some tests but did not outperform the classical Gurobi solver, in either runtime or solution quality, on a power-generation scheduling case [3]. A shared library changes the terms of the debate: claims can be compared on the same instances, with the same metrics, against the best known classical results.
Position in the Civilizational Compute Stack
L3, hybrid workflow. This is the layer where physical progress is converted, or fails to be converted, into economic value, and therefore where measurement discipline matters most.
Why it matters
Institutions, investors and governments are making decisions on the basis of quantum advantage claims. Without common yardsticks, those decisions rest on announcements. With them, advantage becomes a contest that can be scored on total cost: time, solution quality and the strength of the classical alternative. QCTM adopts this principle for every workflow-level indicator.
The open question
Which QOBLIB problem class will be the first where a quantum or hybrid workflow beats the strongest classical baseline on total cost, and does that advantage survive the next round of classical improvement?
Signposts
Quantum and classical results reported on identical QOBLIB instances; independent replication; total-cost reporting that includes queueing, repetitions and classical pre- and post-processing.