Sampling / IQP: which simulator is fastest

Diagonal phase circuit between Hadamard layers

8 simulators ran this circuit at 4 to 24 qubits, 24 workloads in all. Qulacs was the fastest in 12, Google qsim in 10, 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

QubitsPennyLanePennyLane LightningQiskitQiskit AerQiskit Aer (MPS)Google qsimQulacsTensorCircuit
4 qubits
PennyLane
5.68 ms
PennyLane Lightning
2.97 ms
Qiskit
2.20 ms
Qiskit Aer
1.69 ms
Qiskit Aer (MPS)
1.98 ms
Google qsim
2.70 ms
Qulacs
68.8 µs
TensorCircuit
209 µs
8 qubits
PennyLane
11.3 ms
PennyLane Lightning
5.19 ms
Qiskit
4.96 ms
Qiskit Aer
2.99 ms
Qiskit Aer (MPS)
4.58 ms
Google qsim
5.02 ms
Qulacs
156 µs
TensorCircuit
425 µs
12 qubits
PennyLane
19.7 ms
PennyLane Lightning
7.50 ms
Qiskit
11.3 ms
Qiskit Aer
6.66 ms
Qiskit Aer (MPS)
21.5 ms
Google qsim
7.41 ms
Qulacs
1.04 ms
TensorCircuit
1.49 ms
16 qubits
PennyLane
95.5 ms
PennyLane Lightning
15.5 ms
Qiskit
119 ms
Qiskit Aer
31.8 ms
Qiskit Aer (MPS)
61.9 ms
Google qsim
10.8 ms
Qulacs
18.5 ms
TensorCircuit
21.1 ms
20 qubits
PennyLane
2.15 s
PennyLane Lightning
206 ms
Qiskit
3.26 s
Qiskit Aer
540 ms
Qiskit Aer (MPS)
skip
Google qsim
43.8 ms
Qulacs
392 ms
TensorCircuit
633 ms
24 qubits
PennyLane
skip
PennyLane Lightning
7.66 s
Qiskit
skip
Qiskit Aer
10.6 s
Qiskit Aer (MPS)
skip
Google qsim
1.11 s
Qulacs
10.7 s
TensorCircuit
13.7 s

Expectation value, global observable (all_z), depth 3

Time per run, by qubit count

QubitsPennyLanePennyLane LightningQiskitQiskit AerQiskit Aer (MPS)Google qsimQulacsTensorCircuit
4 qubits
PennyLane
6.14 ms
PennyLane Lightning
3.18 ms
Qiskit
2.33 ms
Qiskit Aer
1.79 ms
Qiskit Aer (MPS)
2.11 ms
Google qsim
2.73 ms
Qulacs
78.2 µs
TensorCircuit
229 µs
8 qubits
PennyLane
12.6 ms
PennyLane Lightning
5.57 ms
Qiskit
5.29 ms
Qiskit Aer
3.21 ms
Qiskit Aer (MPS)
5.12 ms
Google qsim
5.21 ms
Qulacs
203 µs
TensorCircuit
464 µs
12 qubits
PennyLane
21.6 ms
PennyLane Lightning
8.14 ms
Qiskit
11.7 ms
Qiskit Aer
7.28 ms
Qiskit Aer (MPS)
36.1 ms
Google qsim
7.69 ms
Qulacs
1.97 ms
TensorCircuit
2.04 ms
16 qubits
PennyLane
105 ms
PennyLane Lightning
20.3 ms
Qiskit
121 ms
Qiskit Aer
39.6 ms
Qiskit Aer (MPS)
167 ms
Google qsim
11.8 ms
Qulacs
38.7 ms
TensorCircuit
59.5 ms
20 qubits
PennyLane
2.48 s
PennyLane Lightning
317 ms
Qiskit
3.09 s
Qiskit Aer
681 ms
Qiskit Aer (MPS)
skip
Google qsim
55.6 ms
Qulacs
797 ms
TensorCircuit
1.53 s
24 qubits
PennyLane
skip
PennyLane Lightning
13.6 s
Qiskit
skip
Qiskit Aer
13.5 s
Qiskit Aer (MPS)
skip
Google qsim
1.50 s
Qulacs
19.1 s
TensorCircuit
45.0 s

Expectation value, global observable (pauli_mix), depth 3

Time per run, by qubit count

QubitsPennyLanePennyLane LightningQiskitQiskit AerQiskit Aer (MPS)Google qsimQulacsTensorCircuit
8 qubits
PennyLane
13.7 ms
PennyLane Lightning
5.87 ms
Qiskit
6.21 ms
Qiskit Aer
3.22 ms
Qiskit Aer (MPS)
5.95 ms
Google qsim
5.64 ms
Qulacs
239 µs
TensorCircuit
568 µs
12 qubits
PennyLane
24.5 ms
PennyLane Lightning
8.96 ms
Qiskit
13.8 ms
Qiskit Aer
7.83 ms
Qiskit Aer (MPS)
65.2 ms
Google qsim
8.40 ms
Qulacs
2.67 ms
TensorCircuit
2.94 ms
16 qubits
PennyLane
127 ms
PennyLane Lightning
28.1 ms
Qiskit
138 ms
Qiskit Aer
49.3 ms
Qiskit Aer (MPS)
359 ms
Google qsim
15.6 ms
Qulacs
53.4 ms
TensorCircuit
96.1 ms
20 qubits
PennyLane
2.74 s
PennyLane Lightning
566 ms
Qiskit
3.26 s
Qiskit Aer
873 ms
Qiskit Aer (MPS)
skip
Google qsim
147 ms
Qulacs
1.09 s
TensorCircuit
2.76 s

Expectation value, global observable (zz_sum), depth 3

Time per run, by qubit count

QubitsPennyLanePennyLane LightningQiskitQiskit AerQiskit Aer (MPS)Google qsimQulacsTensorCircuit
8 qubits
PennyLane
12.8 ms
PennyLane Lightning
5.76 ms
Qiskit
6.01 ms
Qiskit Aer
3.20 ms
Qiskit Aer (MPS)
5.31 ms
Google qsim
5.43 ms
Qulacs
196 µs
TensorCircuit
525 µs
12 qubits
PennyLane
22.2 ms
PennyLane Lightning
8.48 ms
Qiskit
13.0 ms
Qiskit Aer
7.36 ms
Qiskit Aer (MPS)
39.0 ms
Google qsim
8.08 ms
Qulacs
2.04 ms
TensorCircuit
2.56 ms
16 qubits
PennyLane
108 ms
PennyLane Lightning
22.2 ms
Qiskit
134 ms
Qiskit Aer
40.6 ms
Qiskit Aer (MPS)
179 ms
Google qsim
13.3 ms
Qulacs
39.9 ms
TensorCircuit
79.0 ms
20 qubits
PennyLane
2.21 s
PennyLane Lightning
363 ms
Qiskit
3.18 s
Qiskit Aer
703 ms
Qiskit Aer (MPS)
skip
Google qsim
88.2 ms
Qulacs
856 ms
TensorCircuit
2.22 s

Gradient, local observable, depth 3

Time per run, by qubit count

QubitsPennyLanePennyLane LightningQiskitQiskit AerQiskit Aer (MPS)Google qsimQulacsTensorCircuit
4 qubits
PennyLane
20.1 ms
PennyLane Lightning
5.95 ms
Qiskit
unsupported
Qiskit Aer
unsupported
Qiskit Aer (MPS)
unsupported
Google qsim
unsupported
Qulacs
272 µs
TensorCircuit
425 µs
8 qubits
PennyLane
43.1 ms
PennyLane Lightning
10.1 ms
Qiskit
unsupported
Qiskit Aer
unsupported
Qiskit Aer (MPS)
unsupported
Google qsim
unsupported
Qulacs
701 µs
TensorCircuit
898 µs
12 qubits
PennyLane
112 ms
PennyLane Lightning
15.4 ms
Qiskit
unsupported
Qiskit Aer
unsupported
Qiskit Aer (MPS)
unsupported
Google qsim
unsupported
Qulacs
5.38 ms
TensorCircuit
4.39 ms
16 qubits
PennyLane
742 ms
PennyLane Lightning
65.3 ms
Qiskit
unsupported
Qiskit Aer
unsupported
Qiskit Aer (MPS)
unsupported
Google qsim
unsupported
Qulacs
105 ms
TensorCircuit
104 ms

Which quantum computers can run it

At depth 3, 1,000 shots

Quantum computerCan run it up toBorderline up toEstimated cost at that size
Can run it up to
8 qubits
Borderline up to
32 qubits
Estimated cost at that size
$200.30–$600.90
via AWS Braket
Can run it up to
20 qubits
Borderline up to
32 qubits
Estimated cost at that size
$200.30–$600.90
via AWS Braket
Can run it up to
4 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
24 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
32 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
24 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
24 qubits
Estimated cost at that size
$0.725–$2.17
via AWS Braket
Can run it up to
16 qubits
Borderline up to
48 qubits
Estimated cost at that size
$1.90–$5.70
via AWS Braket
Can run it up to
8 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.

Hardware figures each carry the source they came from and the date they were measured. The oldest in this set is 626 days old. Costs are bands, not quotes.

Host: Linux-6.8.0-1051-aws-x86_64-with-glibc2.35, 8 cores, threads=1, core-pinned. 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. Every simulator was driven through one common harness, so any harness binding overhead is charged to the simulator rather than hidden.

© 2026 EigenOrbit. All rights reserved.