Using Modular Open Architecture, a Dutch Collaboration Introduces the Tuna-5 Quantum System.
Tuna-5
A system-readiness platform that is essential to furthering national and EU policies towards scalable quantum systems, Tuna-5 is more than just the introduction of a new quantum processor. It is an important component of the Netherlands’ HectoQubit/2 (HQ/2) project, which was started in 2023 as a project supported by Quantum Delta NL and the National Growth Fund to connect university research and startup-driven innovation. It represents the culmination of contributions from all aspects of the Dutch quantum supply chain.
A 100-qubit demonstrator is the goal of the EU Quantum Flagship project OpenSuperQPlus, which Tuna-5 supports alignment with by the end of 2026.
Tuna-5 is an ecosystem-level approach to co-development, in contrast to vertically integrated quantum computers. The modular architecture incorporates a Python-based SDK, a quantum operating system, a modular control stack, a quantum processor with tunable couplers, and public cloud access. Users can now access the integrated, functional system through the Quantum Inspire platform to the thorough integration and iteration of components from several manufacturers. The Dutch quantum supply chain’s maturity and interoperability are thought to be improved by this cooperative strategy.
Tuna-5 architecture
Every startup that took part added a unique subsystem to Tuna-5’s tiered architecture:
- The manufacturing of the quantum chip, which supplied the processor with tunable couplers, was handled by QuantWare.
- The control electronics required for readout and gate operations were provided by Qblox.
- The Orange Juice operating system and quantum toolkit were supplied by Orange Quantum Systems.
- Although Delft Circuits’ cryogenic cabling is not incorporated into Tuna-5, it is incorporated into a different, scaled-up prototype that is presently being developed.
- TNO coordinates the execution of Python-based quantum algorithms into the required pulse sequences for quantum gate operations, managing the entire software stack.
- The employment of flux-tunable couplers in the processor of Tuna-5 is a significant technological achievement. Compared to the fixed-frequency couplers used in earlier CPU generations, such as the Starmon-5, this is an improvement.
The dynamic modulation of interaction strengths between qubit pairs is made possible by tunable couplers, which is essential for enhancing gate fidelity and lowering errors frequently linked to residual couplings. Recent experimental results from QuTech’s DiCarlo Lab, where Tuna-5 is housed, served as the basis for the system calibration. In order to minimize “spectator effects,” or unexpected interactions from inactive qubits, by carefully nulling certain residual couplings during operations, this study used systematic trials to calibrate and optimize coupler frequencies for a variety of quantum operations.
Best practices for using tunable-coupler processors in noise-sensitive applications like quantum error correction are directly influenced by the findings of this study. The research conducted by the DiCarlo Lab has improved experimental infrastructure and impacted product advancements in the participating companies.
How does Tuna-5 Works
In terms of operation, Tuna-5 is made up of five qubits (Q0-Q4) positioned in a starfish pattern, with Q2 in the centre. A particular set of single- and two-qubit gates are supported by the system. Gates like Rx(angle), Ry(angle), Rz(angle), X90, Y90, mX90, mY90, X, Y, Z, I, S, Sdag, T, Tdag, and CZ are examples of native operations, which only require one instruction on the control hardware. Non-unitary instructions like init, measure, and reset are other examples of native operations.
Although supported, non-native operations such as H, CNOT, SWAP, CR, and CRk are decomposed using processor-specific rules. Binary regulated operations (c-), Display, Display_binary, Not, and Toffoli are among the commands and operations that are specifically prohibited.
QuTech’s OpenSquirrel is used to compile user programs provided to Tuna-5, which includes scheduling and gate decomposition processes. Hardware limitations and operation durations are taken into consideration when scheduling using the as-late-as-possible (ALAP) technique (20 ns for single-qubit operations, 60 ns for two-qubit operations, 800 ns for measurements, and 500 µs for initialization, for example).
Wait and barrier instructions allow users to control scheduling; the former delays further instructions on designated qubits, while the latter stops scheduling over the barrier, establishing a scheduling domain. During compilation, non-native gate operations are broken down into native operations, which impacts how long they take overall. Interestingly, arbitrary rotations with a single qubit are quantized to the closest multiple of π/36 radians (5°).
Measurement results are shown as a 2-dimensional array and are dependent on the user’s program structure. The second dimension encodes qubit measurement results in the order given in the traditional bit register, whereas the outer dimension represents several shots. The mapping of qubit measurements to classical bits is explicitly controlled by the user. Since measurements are susceptible to scheduling, users should use barrier instructions to guarantee that measurements on multiple qubits are carried out in parallel.
Along with the QX emulator and other real-hardware backends like Starmon-7 and Spin-2+, Tuna-5 is now accessible to consumers via the Quantum Inspire platform. Its release comes after Quantum Inspire 2.0, which was released in February 2025 and improved the platform’s execution engine and user interface.
The public release of Tuna-5 highlights the Netherlands’ commitment to open collaboration, modular design, and industry-academic synergy in achieving practical quantum computing, even though it is currently a system-readiness benchmark and not meant as a production-scale machine. In order to expand on the architectural and integration lessons learnt from Tuna-5, future scaled prototypes are already in the development stage.
