June 09, 2026 -- Researchers from Japan have developed quantum multi-programming auto mode, a function that automatically runs quantum programs from different users in parallel. Launched on Japan’s leading research institute, the Center for Quantum Information and Quantum Biology (QIQB)’s quantum computer cloud service, the system reduces idle qubit resources, improves throughput, and may help ease congestion in quantum cloud computing.

Quantum computers are expected to become powerful next-generation computing platforms, but access to real quantum hardware remains limited. Because quantum chips require specialized facilities and careful control to operate stably, many universities, research institutes, and companies now provide access to them through cloud services.

However, cloud access creates a new challenge: waiting. Current noisy intermediate-scale quantum computers have a limited number of qubits, and conventional cloud operation often allows one job to occupy an entire quantum chip. This means that many qubits can remain unused, even while other users are waiting for their programs to run.

A research group led by QIQB at The University of Osaka in collaboration with Systems Engineering Consultants Co.,LTD. (SEC) and Juntendo University, has now developed and launched quantum multi-programming (auto mode), a function that automatically executes quantum programs from different users in parallel on QIQB’s quantum computer cloud service.

The University of Osaka’s quantum computer cloud service operates a 64-qubit quantum chip. However, many research programs use only around 10 qubits. Under the conventional “one job occupies the whole chip” approach, this leaves much of the chip idle.

The new auto mode helps solve this problem by selecting suitable jobs from the cloud queue, assigning them to available qubits, and executing them together. It improves on the previously released quantum multi-programming (manual mode), which allowed parallel execution only for multiple programs manually specified by the same user.

“Reducing computation wait times is one of the key challenges toward practical quantum computing,” says MORI Toshio, Specially Appointed Researcher (full-time) at The University of Osaka. “As the number of qubits in quantum computers continues to grow, we expect demand for quantum multi-programming auto mode to increase. This function can also be used on systems that have not yet adopted OQTOPUS by introducing OQTOPUS as middleware, and we hope to accelerate its deployment to other systems going forward.”

The new function does more than simply fill empty space on a chip. It uses mathematical optimization to determine how multiple quantum circuits can be placed efficiently.

First, the system represents both quantum circuits and the quantum chip as graphs made up of vertices and edges. It then solves a subgraph isomorphism problem, fitting multiple circuit graphs into the chip graph like pieces of a puzzle. By using an integer programming solver, the system can determine placements quickly and accurately, even for complex quantum circuits.

Second, the system automatically accounts for hardware-specific constraints, such as the direction of qubit connections and limited connectivity between distant qubits. The circuits are transformed, or transpiled, before being combined, allowing users to run their programs without needing to manage these physical constraints themselves.

Third, the system is designed with fairness in mind. It checks a fixed number of jobs at the front of the queue and searches for combinations that can be executed in parallel. This allows jobs that have waited longer to be prioritized while still improving overall efficiency.

To test the system, the team used a dataset reflecting real-world user behavior. In one evaluation, five users submitted 110 jobs for two-qubit circuits on an 11-qubit chip, assuming a situation in which small-scale quantum circuits are frequently used for research. The system improved throughput, or the amount of computation processed per unit time, by approximately 3.76 times.

“Maximizing the utilization of qubit resources is a key challenge for the future use of quantum computers,” says UCHIDA Ryo, Technical Manager at SEC. “Quantum multi-programming auto mode reduces wait times and improves resource efficiency through parallel execution of quantum programs and their optimized qubit allocation. We will continue our R&D efforts to advance quantum computing.”

The results suggest that automatic quantum multi-programming can help ease congestion in quantum computer cloud services. By running more quantum circuits at the same time, the system reduces idle qubit resources and improves the operating efficiency of valuable quantum computing infrastructure.

“This work shows that classical optimization methods can contribute to the operation of quantum computers,” says NAKADA Hidemoto, Professor at Juntendo University. “We will continue to explore similar contributions.”

The function has been implemented in OQTOPUS, an open-source basic software stack for quantum computers, and will be provided sequentially to organizations participating in the Quantum Software Consortium that use The University of Osaka’s quantum computer cloud service.

QIQB, SEC, and Juntendo University will continue to advance system software research and development to improve the usability and performance of quantum computers and contribute to the practical application of quantum technologies.