Hardware-efficient VQE: which simulator is fastest
RY/RZ layers with a CNOT ring
8 simulators ran this circuit at 4 to 24 qubits, 19 workloads in all. Qulacs was the fastest in 10, Google qsim in 5, Qiskit Aer (MPS) in 2, TensorCircuit in 1 and PennyLane Lightning in 1. The tables below hold every time as it was measured, with the fastest in each row marked. Open this circuit in the interactive tool to see the hardware that could run it and what a run would cost.
Expectation value, local observable, depth 3
Time per run, by qubit count
| Qubits | PennyLane | PennyLane Lightning | Qiskit | Qiskit Aer | Qiskit Aer (MPS) | Google qsim | Qulacs | TensorCircuit |
|---|---|---|---|---|---|---|---|---|
4 qubits | PennyLane 3.43 ms | PennyLane Lightning 2.18 ms | Qiskit 1.61 ms | Qiskit Aer 1.31 ms | Qiskit Aer (MPS) 1.58 ms | Google qsim 2.21 ms | Qulacs 52.5 µs | TensorCircuit 211 µs |
8 qubits | PennyLane 6.83 ms | PennyLane Lightning 3.48 ms | Qiskit 3.51 ms | Qiskit Aer 2.20 ms | Qiskit Aer (MPS) 3.55 ms | Google qsim 4.06 ms | Qulacs 126 µs | TensorCircuit 386 µs |
12 qubits | PennyLane 12.0 ms | PennyLane Lightning 4.96 ms | Qiskit 7.81 ms | Qiskit Aer 4.69 ms | Qiskit Aer (MPS) 18.4 ms | Google qsim 5.87 ms | Qulacs 1.03 ms | TensorCircuit 1.50 ms |
16 qubits | PennyLane 61.6 ms | PennyLane Lightning 10.3 ms | Qiskit 74.9 ms | Qiskit Aer 40.5 ms | Qiskit Aer (MPS) 57.4 ms | Google qsim 8.94 ms | Qulacs 19.2 ms | TensorCircuit 20.7 ms |
20 qubits | PennyLane 1.41 s | PennyLane Lightning 136 ms | Qiskit 1.90 s | Qiskit Aer 738 ms | Qiskit Aer (MPS) 96.0 ms | Google qsim 40.4 ms | Qulacs 386 ms | TensorCircuit 488 ms |
24 qubits | PennyLane skip | PennyLane Lightning 4.80 s | Qiskit skip | Qiskit Aer 14.6 s | Qiskit Aer (MPS) 134 ms | Google qsim 1.09 s | Qulacs 8.90 s | TensorCircuit 13.7 s |
Expectation value, global observable (all_z), depth 3
Time per run, by qubit count
| Qubits | PennyLane | PennyLane Lightning | Qiskit | Qiskit Aer | Qiskit Aer (MPS) | Google qsim | Qulacs | TensorCircuit |
|---|---|---|---|---|---|---|---|---|
4 qubits | PennyLane 3.91 ms | PennyLane Lightning 2.41 ms | Qiskit 1.80 ms | Qiskit Aer 1.39 ms | Qiskit Aer (MPS) 1.69 ms | Google qsim 2.28 ms | Qulacs 63.7 µs | TensorCircuit 212 µs |
8 qubits | PennyLane 7.93 ms | PennyLane Lightning 3.81 ms | Qiskit 3.83 ms | Qiskit Aer 2.41 ms | Qiskit Aer (MPS) 4.12 ms | Google qsim 4.17 ms | Qulacs 172 µs | TensorCircuit 454 µs |
12 qubits | PennyLane 14.0 ms | PennyLane Lightning 5.56 ms | Qiskit 8.17 ms | Qiskit Aer 5.23 ms | Qiskit Aer (MPS) 34.1 ms | Google qsim 6.29 ms | Qulacs 1.97 ms | TensorCircuit 2.02 ms |
16 qubits | PennyLane 72.0 ms | PennyLane Lightning 14.5 ms | Qiskit 79.1 ms | Qiskit Aer 48.2 ms | Qiskit Aer (MPS) 169 ms | Google qsim 10.0 ms | Qulacs 40.2 ms | TensorCircuit 53.9 ms |
20 qubits | PennyLane 1.54 s | PennyLane Lightning 248 ms | Qiskit 2.04 s | Qiskit Aer 899 ms | Qiskit Aer (MPS) 350 ms | Google qsim 52.9 ms | Qulacs 808 ms | TensorCircuit 1.09 s |
24 qubits | PennyLane skip | PennyLane Lightning 10.6 s | Qiskit 134 s | Qiskit Aer 17.6 s | Qiskit Aer (MPS) 578 ms | Google qsim 1.51 s | Qulacs 17.1 s | TensorCircuit 27.5 s |
Gradient, local observable, depth 3
Time per run, by qubit count
| Qubits | PennyLane | PennyLane Lightning | Qiskit | Qiskit Aer | Qiskit Aer (MPS) | Google qsim | Qulacs | TensorCircuit |
|---|---|---|---|---|---|---|---|---|
4 qubits | PennyLane 19.4 ms | PennyLane Lightning 4.69 ms | Qiskit unsupported | Qiskit Aer unsupported | Qiskit Aer (MPS) unsupported | Google qsim unsupported | Qulacs 206 µs | TensorCircuit 340 µs |
8 qubits | PennyLane 38.7 ms | PennyLane Lightning 7.64 ms | Qiskit unsupported | Qiskit Aer unsupported | Qiskit Aer (MPS) unsupported | Google qsim unsupported | Qulacs 577 µs | TensorCircuit 830 µs |
12 qubits | PennyLane 82.6 ms | PennyLane Lightning 11.5 ms | Qiskit unsupported | Qiskit Aer unsupported | Qiskit Aer (MPS) unsupported | Google qsim unsupported | Qulacs 5.51 ms | TensorCircuit 4.05 ms |
16 qubits | PennyLane 462 ms | PennyLane Lightning 54.3 ms | Qiskit unsupported | Qiskit Aer unsupported | Qiskit Aer (MPS) unsupported | Google qsim unsupported | Qulacs 110 ms | TensorCircuit 66.6 ms |
Statevector, global observable (all_z), depth 3
Time per run, by qubit count
| Qubits | PennyLane | PennyLane Lightning | Qiskit | Qiskit Aer | Qiskit Aer (MPS) | Google qsim | Qulacs | TensorCircuit |
|---|---|---|---|---|---|---|---|---|
8 qubits | PennyLane 6.89 ms | PennyLane Lightning 5.38 ms | Qiskit 3.68 ms | Qiskit Aer 2.31 ms | Qiskit Aer (MPS) 3.85 ms | Google qsim 4.13 ms | Qulacs 129 µs | TensorCircuit unsupported |
12 qubits | PennyLane 12.1 ms | PennyLane Lightning 6.94 ms | Qiskit 7.98 ms | Qiskit Aer 4.81 ms | Qiskit Aer (MPS) 20.7 ms | Google qsim 5.93 ms | Qulacs 1.00 ms | TensorCircuit unsupported |
16 qubits | PennyLane 62.8 ms | PennyLane Lightning 12.4 ms | Qiskit 77.5 ms | Qiskit Aer 42.0 ms | Qiskit Aer (MPS) 125 ms | Google qsim 9.69 ms | Qulacs 18.9 ms | TensorCircuit unsupported |
Which quantum computers can run it
At depth 3, 1,000 shots
| Quantum computer | Can run it up to | Borderline up to | Estimated cost at that size |
|---|---|---|---|
| Can run it up to 16 qubits | Borderline up to 32 qubits | Estimated cost at that size $200.30–$600.90 via AWS Braket | |
| Can run it up to 32 qubits | Borderline up to – | Estimated cost at that size $200.30–$600.90 via AWS Braket | |
| Can run it up to 8 qubits | Borderline up to 12 qubits | Estimated cost at that size $23.80–$71.40 via AWS Braket | |
| Can run it up to 96 qubits | Borderline up to – | Estimated cost at that size Not priced this vendor publishes no rate card | |
| Can run it up to 48 qubits | Borderline up to – | Estimated cost at that size Not priced this vendor publishes no rate card | |
| Can run it up to 48 qubits | Borderline up to – | Estimated cost at that size Not priced this vendor publishes no rate card | |
| Can run it up to – | Borderline up to – | Estimated cost at that size – | |
| Can run it up to – | Borderline up to – | Estimated cost at that size – | |
| Can run it up to – | Borderline up to – | Estimated cost at that size – | |
| Can run it up to 32 qubits | Borderline up to 96 qubits | Estimated cost at that size from $0.015 via IBM Quantum. 0.01s at maximum throughput, so this is a floor | |
| Can run it up to 48 qubits | Borderline up to 128 qubits | Estimated cost at that size from $0.421 via IBM Quantum. 0.26s at maximum throughput, so this is a floor | |
| Can run it up to 32 qubits | Borderline up to 96 qubits | Estimated cost at that size from $0.421 via IBM Quantum. 0.26s at maximum throughput, so this is a floor | |
| Can run it up to 4 qubits | Borderline up to 32 qubits | Estimated cost at that size $0.725–$2.17 via AWS Braket | |
| Can run it up to 24 qubits | Borderline up to 48 qubits | Estimated cost at that size $1.90–$5.70 via AWS Braket | |
| Can run it up to 16 qubits | Borderline up to 20 qubits | Estimated cost at that size $1.75–$5.25 via AWS Braket |
Sizes are the grid's own: 4, 8, 12, 16, 20, 24, 32, 48, 64, 96, 128, 156 qubits. A device can run a circuit when a clean run is more likely than not, and is borderline when it is better than one in fifty. The cost is for one run of 1,000 shots at the largest size in the row, as a band and not a quote.
The simulators
Each one across every circuit family
How this was measured
One host, one harness
Every simulator ran the same circuits on one host, single-threaded and core-pinned, driven through one common harness, so any overhead the harness adds is charged to the simulator. Each result was checked against a reference implementation: a simulator that ran fast but returned the wrong answer is left out of the ranking. Nothing is interpolated or rescaled. A time was measured, or it is not shown.
Run 2026-08-10, 8 cores, threads=1. numpy 2.4.6, scipy 1.18.0, torch 2.13.0, numba 0.66.0, pennylane 0.45.1, pennylane_lightning 0.45.0, qiskit 2.5.1, qulacs 0.6.14, tensorcircuit 1.8.0, cudaq 0.15.1. One host and one thread make the comparison reproducible. It is not a prediction for your laptop.
The interactive tool shows the same measurements for any workload you pick, with the quantum hardware that could run it and what a run would cost.