IBM launches cloud-native, pay-as-you-go service for quantum computing

On April 12, IBM Cloud launched a beta version of its new pay-as-you-go program, giving anyone with an IBM Cloud account access to two 27-qubit Falcon processors. Users can develop demanding quantum programs using Qiskit runtime (Runtime) primitives and execute them seamlessly on IBM's state-of-the-art systems, simply using a credit card or IBM cloud credits to purchase the required resources for a fee of $1.60 /runtime second.

Offering the service through the IBM Cloud opens the door for IBM's ecosystem partners who build on the Qiskit runtime API and use the IBM Cloud's user management, billing and other supporting partner infrastructure, IBM said .

Strangeworks also announced that it is the first IBM Business Partner to offer consumption-based pricing access to select IBM Quantum systems through the Qiskit runtime's pay-as-you-go pricing model.

A year ago, IBM introduced the Qiskit runtime, which allows users to bundle quantum and classical execution into programs that run on classical computing infrastructure co-located with IBM Quantum systems. By changing the execution model from services executing circuits to services executing programs, some of the biggest performance bottlenecks affecting workload performance were successfully removed.

In May 2021, IBM Quantum achieved 120x quantum workload acceleration with Qiskit runtime.

Now, to further simplify the developer experience, IBM has built two Qiskit runtime primitives, Sampler and Estimator. The core ability of quantum computers to distinguish them from classical computers is their ability to generate non-classical probability distributions at the output. The native operation one can do with a probability distribution is to sample or estimate quantities from it. Thus, these sampling and estimation operations form a fundamental part of the development of quantum algorithms.

IBM's first two Qiskit runtime primitives directly expose these sampling and estimation operations as core interfaces to its quantum systems via samplers and estimators, respectively.

1. Sampler

The sampler estimates the entire quasi-probability distribution of the quantum circuit's output by sampling from the quantum circuit's output. This is useful for search algorithms like Grover search.

2. Estimator

The estimator computes the expected value of the observable at the output of the circuit. Such observables can encode many things, such as the electronic structure of molecules, cost functions for optimization problems, kernels for machine learning problems, and more.

When developers use these primitives, they want to express their operational needs at the same time; that is, developers may need to know what to expect with a specific target precision or with a maximum execution time. These goals are distinct from trying to control low-level details, such as the number of repetitions or the specific error mitigation method used to achieve the target accuracy.

Compared to any other quantum service, the Qiskit runtime primitives provide an experience that is faster, more efficient, and more tailored to the needs of algorithm developers. The efficiency and speed gains of the Qiskit runtime are inherently scalable. With new optimizations to the platform, programs run faster and developers have less work to do. Over time, tools such as bug suppression and mitigation will be introduced into the service to further enhance its capabilities.

In the future, developers can quickly and easily create programs to solve those problems by simply asking questions, and executing those programs seamlessly, without requiring any hardware configuration. IBM wants to enable a frictionless quantum future with the help of the Qiskit runtime.

Link:

[1] https://research.ibm.com/blog/qiskit-runtime-for-useful-quantum-computing

[2] https://www.ibm.com/cloud/blog/how-to-make-quantum-a-pay-as-you-go-cloud-service

[3] https://www.newswire.com/news/strangeworks-announces-availability-of-consumption-pricing-model-for-21683409