An international consortium led by IBM Quantum, the Zuse Institute Berlin (ZIB), Technische Universitat Berlin and Purdue University on August 15, 2026, launched QOBLIB — the Quantum Optimization Benchmarking Library — an open-source suite designed to give the industry a common yardstick for measuring progress toward practical quantum advantage.
Ten NP-Hard Problems, One Scoreboard
QOBLIB packages ten NP-hard combinatorial problem classes — from maximum independent set and QUBO variants to graph coloring — into a standardized, model-independent framework that scores quantum, classical and hybrid solvers head-to-head. A public web portal tracks live results and lets solvers submit runs to keep pace with state-of-the-art classical baselines.
Why It Matters
Quantum optimization has been the loudest early candidate for near-term quantum utility, but claims have been hard to compare across hardware and heuristics. By standardising the problem set and metrics, QOBLIB gives buyers a way to check vendor claims and gives researchers a shared regression test — a similar role to MLPerf in classical AI. Lead authors Thorsten Koch (ZIB/TU Berlin) and Stefan Woerner (IBM Quantum) published the framework in Nature Computational Science with more than 2,000 submitted results at launch.
Consortium And Roadmap
Beyond the four founders, the QOBLIB consortium spans other academic and industrial partners, and slots alongside IBM's broader push — including Quanta Computer's industrial quantum tie-ups and Quantinuum-Oracle Helios integration. The GitHub repository (ZIB-AOPT/QOBLIB) is live and accepting community submissions.
Reporting based on coverage from IBM Research, Quantum Computing Report and Quantum Zeitgeist.