Which quantum backend should run your circuit? Speed, feasibility and cost in one table

dev.to Python2026-11-13 21:460 views

Before a circuit goes anywhere, three questions come up. How long will it take on each simulator? Will it fit on a given device at all? What will it cost on each QPU?

The answers sit in different places: simulator docs, device spec sheets, pricing pages that each count in their own units (per shot, per task, per second). Putting them side by side usually means a spreadsheet you rebuild every time a price or a device changes.

EigenOrbit Benchmarks puts them on one page. One workload, measured across eight simulators and priced across fourteen QPUs. Every number carries the source it came from.

Speed: measured, on eight simulators

The simulator numbers are runs of the same circuit through eigenkit, which executes one codebase on every backend. Same circuit, same parameters, so the difference you see is the simulator.

This matters more than it sounds. In our paper comparing 12 backends, no single simulator was consistently fastest: the ranking changed with circuit size, structure and batching. A table for your own circuit shape is more useful than a general leaderboard.

Feasibility: will it run there at all

Before cost, a device has to accept the circuit. The page shows, per QPU, whether the workload fits: qubit count, connectivity and supported gates.

On hardware, the paper found that the better processor for a given algorithm depends on how closely the circuit's topology matches the device's physical connectivity. A device with more qubits is not automatically the better choice.

Cost: priced on fourteen QPUs

Each provider prices differently. The page converts all of them into the price of running this workload, so the fourteen QPUs can be compared directly. Each price links to the provider's own pricing page it was taken from.

Every number has a source

Measured runtimes point to the run that produced them. Prices and device specs point to the provider page they were taken from, with the date. If a number looks wrong, you can check it in one click.

From the table to a real run

Once you have picked a device, the same circuit runs there through EigenOrbit Router: one key for all supported vendors, a cost estimate before the job starts, and a prepaid balance so a job you cannot afford does not start. The walkthrough is in the previous post.

If you are still deciding whether quantum fits your problem in the first place, Adviser starts one step earlier.