Quantuva is the software layer for quantum computing — a high-level SDK, a circuit compiler, and error-mitigation tooling that let researchers and algorithm developers work in |ψ⟩, not in gate tables.
Quantum hardware moves fast and breaks API contracts. Quantuva sits between your algorithm and the hardware, so your research survives the next generation of chips.
Write algorithms as readable circuits and observables — not raw gate lists. First-class states, measurements, and noise models in Python.
A compiler that routes, synthesizes, and schedules your circuit for a specific backend — cutting depth without changing the answer.
Noise-aware readout and gate-error mitigation, applied transparently — so the number you report is closer to the number you meant to measure.
You describe the physics you care about. Quantuva handles the part that has nothing to do with your science — transpiling, noise, and the moving target of hardware interfaces.
from quantuva import Circuit, Backend, Observable qc = Circuit(4) qc.h(0) for i in range(3): qc.cx(0, i + 1) qc.measure_all() backend = Backend("superconducting-v2") compiled = qv.compile(qc, backend=backend, opt_level=2) result = qv.run(compiled, shots=2048) obs = Observable("Z(0) * Z(1)") value = qv.expectation(result, obs, mitigate="zne") print(value) # 0.912 ± 0.004
Target the hardware you have access to — or none at all. Quantuva's simulator is fast enough to be your day-to-day, and the same code path deploys to real devices when you need it.
A GPU-accelerated simulator for prototyping and verifying circuits before you spend hardware time.
Model decoherence and readout error locally, so mitigation strategies can be tested offline.
Compile and submit to partner hardware with topology-aware routing and native gate synthesis.
For shallow, high-qubit circuits, matrix-product-state methods extend what you can classically reach.
On matching simulation step size to the observable you actually measure, not the one you can compute.
2026-07-02How few mitigation points you really need before extrapolation stops being worth the overhead.
2026-05-27Estimating success probability without a full reference-state reconstruction.
“We had three different toolchains for three different backends. Quantuva let us write the algorithm once and stop re-implementing the plumbing every time a new device arrived.”