Calibration of the Quantum Computers

Published

April 29, 2026

Q50 and Q5 are continuously calibrated to ensure functionality. The calibration process involves a series of experiments, aimed at fine-tuning the parameters, necessary for controlling and measuring the qubits. In addition to calibration, we run benchmarks to obtain the figures of merit reflecting the current state of the quantum computer.

Calibration sequences

To minimize the impact of calibration on user operations, we schedule the calibration of the quantum computers twice a day at 00:45 UTC and 14:15 UTC, by running a specific sequence of experiments in order. Calibration and benchmarking experiments are interleaved with regular user jobs in device’s job queue. The calibration will affect the waiting time in the queue of regular user jobs at that time.

Calibration:

Twice a day: at 00:45 UTC and 14:15 UTC
Adjusts qubit drive frequency, amplitude fine-tuning and readout threshold
Recalibrates CZ Gate, Virtual Z rotations and performs randomized benchmarking.
Measures T_1, T_2, T_2^*, readout accuracy, single- and two-qubit gate fidelities

A calibration sequence produces what is called a calibration_set. This is a set of device parameters, which the quantum computer is currently using to execute quantum circuits. It is identified via a calibration_set_id, a unique identifier for the specific calibration_set. Usually, when submitting quantum circuits, the most up-to-date calibration set is used, however, it is possible to use a specific calibration_set_id. This can be useful for testing the degradation of the performance of our quantum computers.

The calibration set sometimes contains observations with the suffix par=d2. This refers to the distance of the qubit pairs that can run in parallel, which is necessary to avoid neighbouring qubit pairs of the same group.

Quality metrics set

The benchmarks results are summarized in the quality metric set representing the figures of merit and reflecting the latest state of the quantum computer. The calibration metrics can be fetched from QX’s API by sending HTTP GET requests to /calibration/metrics/latest endpoint corresponding to each device. Each quality metrics set is identified via a unique ID, with a new ID created after each calibration. Note, that not all metrics are updated after each calibration run. We recommend to save the current calibration set ID together with the job ID to facilitate debugging and analysis. Also note that it can happen that there is a delay of up to 10 minutes between calibrating the device and benchmarking the quality metrics set, causing there to be some delay in presenting the quality metrics data.

The metrics contained in the quality metrics set are summarized below:

Metric	Description
T_1 time (s)	The T_1 time is called the longitudinal relaxation rate and describes how quickly the excited state of the qubit returns to its ground state.
T_2 time (s)	The T_2 time is called the transverse relaxation rate and describes loss of coherence of a superposition state.
T_2-echo time (s)	The T_2-echo time describes the loss of coherence of the superposition state of the qubit. It is more precise than the T_2 Time as it is less susceptible to low-frequency noise.
Single-shot readout fidelity	Measures the average accuracy of distinguishing qubit states. The experiment prepares for 50% of the shots the qubit in the ground state \|0> and for the other 50% in the excited state \|1>.
Single-shot readout 10 error	The error in labelling the qubit state as \|0> when it was prepared in state \|1>
Single-shot readout 01 error	The error in labelling the qubit state as \|1> when it was prepared in state \|0>
1QB average gate fidelity	Average 1QB gate fidelity estimated with randomized benchmarking.
2QB average gate fidelity	The average 2QB gate fidelity estimated with randomized benchmarking.
CZ gate fidelity	The average CZ gate fidelity estimated with interleaved randomized benchmarking. It is usually higher than the average 2QB gate fidelity as a random 2QB Clifford transpiles on average to 8 1QB gates and 1.5 CZ gates.

These metrics provide -critical insights into the operational efficiency, error rates, and coherence properties of the quantum computers. Understanding these figures of merit allows a user to make informed decisions on transpilation strategies and qubit selection, optimizing for circuit depth or fidelity.

Example response

Here is an example response from QX’s API for the latest calibration metrics (GET /api/devices/{device}/calibration/metrics/latest):

{
  "calibration_set_id": "3fa85f64-5717-4562-b3fc-2c963f66afa6",
  "calibration_set_created_timestamp": "2025-01-01T00:00:00Z",
  "calibration_set_end_timestamp": "2025-01-01T01:00:00Z",
  "calibration_set_is_invalid": false,
  "metrics": {
    "t1_time": {
      "QB1": { "value": "0.000123", "unit": "s", "uncertainty": "0.000005", "timestamp": "2025-01-01T00:30:00Z" },
      "QB2": { "value": "0.000098", "unit": "s", "uncertainty": "0.000004", "timestamp": "2025-01-01T00:30:00Z" }
    },
    "t2_time": {
      "QB1": { "value": "0.000056", "unit": "s", "uncertainty": "0.000003", "timestamp": "2025-01-01T00:30:00Z" }
    },
    "t2_echo_time": {},
    "clifford_rb_xy_fidelity": {},
    "cz_irb_tgss_crf_fidelity": {},
    "prx_rb_drag_crf_fidelity": {},
    "clifford_rb_uz_cz_fidelity": {},
    "measure_ssro_constant_fidelity": {},
    "measure_qndness_constant_fidelity": {},
    "measure_qndness_constant_qndness_0": {},
    "measure_qndness_constant_qndness_1": {},
    "measure_ssro_constant_error_0_to_1": {},
    "measure_ssro_constant_error_1_to_0": {}
  }
}

A specific calibration set can be retrieved by its ID: GET /api/devices/{device}/calibration/metrics/{calibration_set_id}.

Calibration history

The history of calibration runs for a device is available at:

GET /api/devices/{device}/calibration/runs

The endpoint returns a paginated list of calibration runs (default 20 per page, max 100). The following query parameters are supported:

Parameter	Description
`start_date`	Filter runs after this date (`YYYY-MM-DDTHH:MM:SSZ`)
`end_date`	Filter runs before this date (`YYYY-MM-DDTHH:MM:SSZ`)
`order`	Sort by creation date: `ascending` or `descending` (default)
`limit`	Maximum number of results to return
`page`	Page number for paginated results
`page_size`	Results per page (default 20, max 100)
`include_metrics`	Include full metrics JSON per run (`true` or `1` to enable)

User limit: Regular users can only access calibration runs from the last 30 days. If an earlier start_date is requested, it is silently capped to 30 days ago.

Example response:

{
  "count": 1,
  "next": null,
  "previous": null,
  "results": [
    {
      "id": "3fa85f64-5717-4562-b3fc-2c963f66afa6",
      "device": "my-device",
      "creation_date": "2025-01-01T00:00:00Z",
      "completed_date": "2025-01-01T01:00:00Z",
      "calibration_set_id": "4fa85f64-5717-4562-b3fc-2c963f66afa6",
      "calibration_set_created_timestamp": "2025-01-01T00:00:00Z",
      "quality_metric_set_id": "5fa85f64-5717-4562-b3fc-2c963f66afa6",
      "successful": true,
      "status": "ready"
    }
  ]
}